update

2025-12-10 15:57:07 +00:00 · 2025-12-10 15:52:41 +00:00 · 2025-12-10 15:50:11 +00:00 · 2025-12-10 12:46:47 +00:00 · 2025-12-10 12:46:17 +00:00 · 2025-12-10 12:43:18 +00:00
49 changed files with 3914 additions and 2877 deletions
--- a/.env.local
+++ b/.env.local
@@ -0,0 +1,9 @@
 # Munich News Daily - Local Development Environment Variables
 # MongoDB Configuration
 MONGO_USERNAME=admin
 MONGO_PASSWORD=local123
 MONGO_AUTH=--auth
 # Ollama Model (use smaller/faster model for local dev)
 OLLAMA_MODEL=phi3:latest
--- a/.gitignore
+++ b/.gitignore
@@ -84,7 +84,9 @@ yarn.lock
 .env.production.local
 *.env
 !.env.example
 !.env.local
 !backend/.env.example
 !backend/.env.local
 # ===================================
 # Database
--- a/.kiro/specs/ai-article-summarization/design.md
+++ b/.kiro/specs/ai-article-summarization/design.md
@@ -1,487 +0,0 @@
 # Design Document - AI Article Summarization
 ## Overview
 This design integrates Ollama AI into the news crawler workflow to automatically generate concise summaries of articles. The system will extract full article content, send it to Ollama for summarization, and store both the original content and the AI-generated summary in MongoDB.
 ## Architecture
 ### High-Level Flow
 ```
 RSS Feed → Extract Content → Summarize with Ollama → Store in MongoDB
                ↓                      ↓                    ↓
         Full Article Text    AI Summary (≤150 words)   Both Stored
 ```
 ### Component Diagram
 ```
 ┌─────────────────────────────────────────────────────────────┐
 │                    News Crawler Service                      │
 │                                                              │
 │  ┌────────────────┐      ┌──────────────────┐             │
 │  │ RSS Parser     │──────→│ Content Extractor│             │
 │  └────────────────┘      └──────────────────┘             │
 │                                   │                         │
 │                                   ↓                         │
 │                          ┌──────────────────┐              │
 │                          │ Ollama Client    │              │
 │                          │ (New Component)  │              │
 │                          └──────────────────┘              │
 │                                   │                         │
 │                                   ↓                         │
 │                          ┌──────────────────┐              │
 │                          │ Database Writer  │              │
 │                          └──────────────────┘              │
 └─────────────────────────────────────────────────────────────┘
                                   │
                                   ↓
                          ┌──────────────────┐
                          │  Ollama Server   │
                          │  (External)      │
                          └──────────────────┘
                                   │
                                   ↓
                          ┌──────────────────┐
                          │    MongoDB       │
                          └──────────────────┘
 ```
 ## Components and Interfaces
 ### 1. Ollama Client Module
 **File:** `news_crawler/ollama_client.py`
 **Purpose:** Handle communication with Ollama server for summarization
 **Interface:**
 ```python
 class OllamaClient:
    def __init__(self, base_url, model, api_key=None, enabled=True):
        """Initialize Ollama client with configuration"""
    def summarize_article(self, content: str, max_words: int = 150) -> dict:
        """
        Summarize article content using Ollama
        Args:
            content: Full article text
            max_words: Maximum words in summary (default 150)
        Returns:
            {
                'summary': str,           # AI-generated summary
                'word_count': int,        # Summary word count
                'success': bool,          # Whether summarization succeeded
                'error': str or None,     # Error message if failed
                'duration': float         # Time taken in seconds
            }
        """
    def is_available(self) -> bool:
        """Check if Ollama server is reachable"""
    def test_connection(self) -> dict:
        """Test connection and return server info"""
 ```
 **Key Methods:**
 1. **summarize_article()**
   - Constructs prompt for Ollama
   - Sends HTTP POST request
   - Handles timeouts and errors
   - Validates response
   - Returns structured result
 2. **is_available()**
   - Quick health check
   - Returns True/False
   - Used before attempting summarization
 3. **test_connection()**
   - Detailed connection test
   - Returns server info and model list
   - Used for diagnostics
 ### 2. Enhanced Crawler Service
 **File:** `news_crawler/crawler_service.py`
 **Changes:**
 ```python
 # Add Ollama client initialization
 from ollama_client import OllamaClient
 # Initialize at module level
 ollama_client = OllamaClient(
    base_url=os.getenv('OLLAMA_BASE_URL'),
    model=os.getenv('OLLAMA_MODEL'),
    api_key=os.getenv('OLLAMA_API_KEY'),
    enabled=os.getenv('OLLAMA_ENABLED', 'false').lower() == 'true'
 )
 # Modify crawl_rss_feed() to include summarization
 def crawl_rss_feed(feed_url, feed_name, max_articles=10):
    # ... existing code ...
    # After extracting content
    article_data = extract_article_content(article_url)
    # NEW: Summarize with Ollama
    summary_result = None
    if ollama_client.enabled and article_data.get('content'):
        print(f"   🤖 Summarizing with AI...")
        summary_result = ollama_client.summarize_article(
            article_data['content'],
            max_words=150
        )
        if summary_result['success']:
            print(f"   ✓ Summary generated ({summary_result['word_count']} words)")
        else:
            print(f"   ⚠ Summarization failed: {summary_result['error']}")
    # Build article document with summary
    article_doc = {
        'title': article_data.get('title'),
        'author': article_data.get('author'),
        'link': article_url,
        'content': article_data.get('content'),
        'summary': summary_result['summary'] if summary_result and summary_result['success'] else None,
        'word_count': article_data.get('word_count'),
        'summary_word_count': summary_result['word_count'] if summary_result and summary_result['success'] else None,
        'source': feed_name,
        'published_at': extract_published_date(entry),
        'crawled_at': article_data.get('crawled_at'),
        'summarized_at': datetime.utcnow() if summary_result and summary_result['success'] else None,
        'created_at': datetime.utcnow()
    }
 ```
 ### 3. Configuration Module
 **File:** `news_crawler/config.py` (new file)
 **Purpose:** Centralize configuration management
 ```python
 import os
 from dotenv import load_dotenv
 load_dotenv(dotenv_path='../.env')
 class Config:
    # MongoDB
    MONGODB_URI = os.getenv('MONGODB_URI', 'mongodb://localhost:27017/')
    DB_NAME = 'munich_news'
    # Ollama
    OLLAMA_BASE_URL = os.getenv('OLLAMA_BASE_URL', 'http://localhost:11434')
    OLLAMA_MODEL = os.getenv('OLLAMA_MODEL', 'phi3:latest')
    OLLAMA_API_KEY = os.getenv('OLLAMA_API_KEY', '')
    OLLAMA_ENABLED = os.getenv('OLLAMA_ENABLED', 'false').lower() == 'true'
    OLLAMA_TIMEOUT = int(os.getenv('OLLAMA_TIMEOUT', '30'))
    # Crawler
    RATE_LIMIT_DELAY = 1  # seconds between requests
    MAX_CONTENT_LENGTH = 50000  # characters
 ```
 ## Data Models
 ### Updated Article Schema
 ```javascript
 {
  _id: ObjectId,
  title: String,
  author: String,
  link: String,                    // Unique index
  content: String,                 // Full article content
  summary: String,                 // AI-generated summary (≤150 words)
  word_count: Number,              // Original content word count
  summary_word_count: Number,      // Summary word count
  source: String,
  published_at: String,
  crawled_at: DateTime,
  summarized_at: DateTime,         // When AI summary was generated
  created_at: DateTime
 }
 ```
 ### Ollama Request Format
 ```json
 {
  "model": "phi3:latest",
  "prompt": "Summarize the following article in 150 words or less. Focus on the key points and main message:\n\n[ARTICLE CONTENT]",
  "stream": false,
  "options": {
    "temperature": 0.7,
    "max_tokens": 200
  }
 }
 ```
 ### Ollama Response Format
 ```json
 {
  "model": "phi3:latest",
  "created_at": "2024-11-10T16:30:00Z",
  "response": "The AI-generated summary text here...",
  "done": true,
  "total_duration": 5000000000
 }
 ```
 ## Error Handling
 ### Error Scenarios and Responses
 | Scenario | Handling | User Impact |
 |----------|----------|-------------|
 | Ollama server down | Log warning, store original content | Article saved without summary |
 | Ollama timeout (>30s) | Cancel request, store original | Article saved without summary |
 | Empty summary returned | Log error, store original | Article saved without summary |
 | Invalid response format | Log error, store original | Article saved without summary |
 | Network error | Retry once, then store original | Article saved without summary |
 | Model not found | Log error, disable Ollama | All articles saved without summaries |
 ### Error Logging Format
 ```python
 {
    'timestamp': datetime.utcnow(),
    'article_url': article_url,
    'error_type': 'timeout|connection|invalid_response|empty_summary',
    'error_message': str(error),
    'ollama_config': {
        'base_url': OLLAMA_BASE_URL,
        'model': OLLAMA_MODEL,
        'enabled': OLLAMA_ENABLED
    }
 }
 ```
 ## Testing Strategy
 ### Unit Tests
 1. **test_ollama_client.py**
   - Test summarization with mock responses
   - Test timeout handling
   - Test error scenarios
   - Test connection checking
 2. **test_crawler_with_ollama.py**
   - Test crawler with Ollama enabled
   - Test crawler with Ollama disabled
   - Test fallback when Ollama fails
   - Test rate limiting
 ### Integration Tests
 1. **test_end_to_end.py**
   - Crawl real RSS feed
   - Summarize with real Ollama
   - Verify database storage
   - Check all fields populated
 ### Manual Testing
 1. Test with Ollama enabled and working
 2. Test with Ollama disabled
 3. Test with Ollama unreachable
 4. Test with slow Ollama responses
 5. Test with various article lengths
 ## Performance Considerations
 ### Timing Estimates
 - Article extraction: 2-5 seconds
 - Ollama summarization: 5-15 seconds (depends on article length and model)
 - Database write: <1 second
 - **Total per article: 8-21 seconds**
 ### Optimization Strategies
 1. **Sequential Processing**
   - Process one article at a time
   - Prevents overwhelming Ollama
   - Easier to debug
 2. **Timeout Management**
   - 30-second timeout per request
   - Prevents hanging on slow responses
 3. **Rate Limiting**
   - 1-second delay between articles
   - Respects server resources
 4. **Future: Batch Processing**
   - Queue articles for summarization
   - Process in batches
   - Use Celery for async processing
 ### Resource Usage
 - **Memory**: ~100MB per crawler instance
 - **Network**: ~1-5KB per article (to Ollama)
 - **Storage**: +150 words per article (~1KB)
 - **CPU**: Minimal (Ollama does the heavy lifting)
 ## Security Considerations
 1. **API Key Storage**
   - Store in environment variables
   - Never commit to git
   - Use secrets management in production
 2. **Content Sanitization**
   - Don't log full article content
   - Sanitize URLs in logs
   - Limit error message detail
 3. **Network Security**
   - Support HTTPS for Ollama
   - Validate SSL certificates
   - Use secure connections
 4. **Rate Limiting**
   - Prevent abuse of Ollama server
   - Implement backoff on errors
   - Monitor usage patterns
 ## Deployment Considerations
 ### Environment Variables
 ```bash
 # Required
 OLLAMA_BASE_URL=http://localhost:11434
 OLLAMA_MODEL=phi3:latest
 OLLAMA_ENABLED=true
 # Optional
 OLLAMA_API_KEY=your-api-key
 OLLAMA_TIMEOUT=30
 ```
 ### Docker Deployment
 ```yaml
 # docker-compose.yml
 services:
  crawler:
    build: ./news_crawler
    environment:
      - OLLAMA_BASE_URL=http://ollama:11434
      - OLLAMA_ENABLED=true
    depends_on:
      - ollama
      - mongodb
  ollama:
    image: ollama/ollama:latest
    ports:
      - "11434:11434"
    volumes:
      - ollama_data:/root/.ollama
 ```
 ### Monitoring
 1. **Metrics to Track**
   - Summarization success rate
   - Average summarization time
   - Ollama server uptime
   - Error frequency by type
 2. **Logging**
   - Log all summarization attempts
   - Log errors with context
   - Log performance metrics
 3. **Alerts**
   - Alert if Ollama is down >5 minutes
   - Alert if success rate <80%
   - Alert if average time >20 seconds
 ## Migration Plan
 ### Phase 1: Add Ollama Client (Week 1)
 - Create ollama_client.py
 - Add configuration
 - Write unit tests
 - Test with sample articles
 ### Phase 2: Integrate with Crawler (Week 1)
 - Modify crawler_service.py
 - Add summarization step
 - Update database schema
 - Test end-to-end
 ### Phase 3: Update Backend API (Week 2)
 - Update news routes
 - Add summary fields to responses
 - Update frontend to display summaries
 - Deploy to production
 ### Phase 4: Monitor and Optimize (Ongoing)
 - Monitor performance
 - Tune prompts for better summaries
 - Optimize rate limiting
 - Add batch processing if needed
 ## Rollback Plan
 If issues arise:
 1. **Immediate**: Set `OLLAMA_ENABLED=false`
 2. **Short-term**: Revert crawler code changes
 3. **Long-term**: Remove Ollama integration
 System will continue to work with original content if Ollama is disabled.
 ## Success Metrics
 - ✅ 95%+ of articles successfully summarized
 - ✅ Average summarization time <15 seconds
 - ✅ Zero data loss (all articles stored even if summarization fails)
 - ✅ Ollama uptime >99%
 - ✅ Summary quality: readable and accurate (manual review)
 ## Future Enhancements
 1. **Multi-language Support**
   - Detect article language
   - Use appropriate model
   - Translate summaries
 2. **Custom Summary Lengths**
   - Allow configuration per feed
   - Support different lengths for different use cases
 3. **Sentiment Analysis**
   - Add sentiment score
   - Categorize as positive/negative/neutral
 4. **Keyword Extraction**
   - Extract key topics
   - Enable better search
 5. **Batch Processing**
   - Queue articles
   - Process in parallel
   - Use Celery for async
 6. **Caching**
   - Cache summaries
   - Avoid re-processing
   - Use Redis for cache
--- a/.kiro/specs/ai-article-summarization/requirements.md
+++ b/.kiro/specs/ai-article-summarization/requirements.md
@@ -1,164 +0,0 @@
 # Requirements Document
 ## Introduction
 This feature integrates Ollama AI into the news crawler to automatically summarize articles before storing them in the database. Instead of storing full article content, the system will generate concise 150-word summaries using AI, making the content more digestible for newsletter readers and reducing storage requirements.
 ## Glossary
 - **Crawler Service**: The standalone microservice that fetches and processes article content from RSS feeds
 - **Ollama Server**: The AI inference server that provides text summarization capabilities
 - **Article Content**: The full text extracted from a news article webpage
 - **Summary**: A concise AI-generated version of the article content (max 150 words)
 - **MongoDB**: The database where articles and summaries are stored
 ## Requirements
 ### Requirement 1: Ollama Integration in Crawler
 **User Story:** As a system administrator, I want the crawler to use Ollama for summarization, so that articles are automatically condensed before storage.
 #### Acceptance Criteria
 1. WHEN the crawler extracts article content, THE Crawler Service SHALL send the content to the Ollama Server for summarization
 2. WHEN sending content to Ollama, THE Crawler Service SHALL include a prompt requesting a summary of 150 words or less
 3. WHEN Ollama returns a summary, THE Crawler Service SHALL validate that the summary is not empty
 4. IF the Ollama Server is unavailable, THEN THE Crawler Service SHALL store the original content without summarization and log a warning
 5. WHEN summarization fails, THE Crawler Service SHALL continue processing other articles without stopping
 ### Requirement 2: Configuration Management
 **User Story:** As a system administrator, I want to configure Ollama settings, so that I can control the summarization behavior.
 #### Acceptance Criteria
 1. THE Crawler Service SHALL read Ollama configuration from environment variables
 2. THE Crawler Service SHALL support the following configuration options:
   - OLLAMA_BASE_URL (server URL)
   - OLLAMA_MODEL (model name)
   - OLLAMA_ENABLED (enable/disable flag)
   - OLLAMA_API_KEY (optional authentication)
 3. WHERE OLLAMA_ENABLED is false, THE Crawler Service SHALL store original content without summarization
 4. WHERE OLLAMA_ENABLED is true AND Ollama is unreachable, THE Crawler Service SHALL log an error and store original content
 ### Requirement 3: Summary Storage
 **User Story:** As a developer, I want summaries stored in the database, so that the frontend can display concise article previews.
 #### Acceptance Criteria
 1. WHEN a summary is generated, THE Crawler Service SHALL store it in the `summary` field in MongoDB
 2. WHEN storing an article, THE Crawler Service SHALL include both the original content and the AI summary
 3. THE Crawler Service SHALL store the following fields:
   - `content` (original full text)
   - `summary` (AI-generated, max 150 words)
   - `word_count` (original content word count)
   - `summary_word_count` (summary word count)
   - `summarized_at` (timestamp when summarized)
 4. WHEN an article already has a summary, THE Crawler Service SHALL not re-summarize it
 ### Requirement 4: Error Handling and Resilience
 **User Story:** As a system administrator, I want the crawler to handle AI failures gracefully, so that the system remains reliable.
 #### Acceptance Criteria
 1. IF Ollama returns an error, THEN THE Crawler Service SHALL log the error and store the original content
 2. IF Ollama times out (>30 seconds), THEN THE Crawler Service SHALL cancel the request and store the original content
 3. IF the summary is empty or invalid, THEN THE Crawler Service SHALL store the original content
 4. WHEN an error occurs, THE Crawler Service SHALL include an error indicator in the database record
 5. THE Crawler Service SHALL continue processing remaining articles after any summarization failure
 ### Requirement 5: Performance and Rate Limiting
 **User Story:** As a system administrator, I want the crawler to respect rate limits, so that it doesn't overwhelm the Ollama server.
 #### Acceptance Criteria
 1. THE Crawler Service SHALL wait at least 1 second between Ollama API calls
 2. THE Crawler Service SHALL set a timeout of 30 seconds for each Ollama request
 3. WHEN processing multiple articles, THE Crawler Service SHALL process them sequentially to avoid overloading Ollama
 4. THE Crawler Service SHALL log the time taken for each summarization
 5. THE Crawler Service SHALL display progress indicators showing summarization status
 ### Requirement 6: Monitoring and Logging
 **User Story:** As a system administrator, I want detailed logs of summarization activity, so that I can monitor and troubleshoot the system.
 #### Acceptance Criteria
 1. THE Crawler Service SHALL log when summarization starts for each article
 2. THE Crawler Service SHALL log the original word count and summary word count
 3. THE Crawler Service SHALL log any errors or warnings from Ollama
 4. THE Crawler Service SHALL display a summary of total articles summarized at the end
 5. THE Crawler Service SHALL include summarization statistics in the final report
 ### Requirement 7: API Endpoint Updates
 **User Story:** As a frontend developer, I want API endpoints to return summaries, so that I can display them to users.
 #### Acceptance Criteria
 1. WHEN fetching articles via GET /api/news, THE Backend API SHALL include the `summary` field if available
 2. WHEN fetching a single article via GET /api/news/<url>, THE Backend API SHALL include both `content` and `summary`
 3. THE Backend API SHALL include a `has_summary` boolean field indicating if AI summarization was performed
 4. THE Backend API SHALL include `summarized_at` timestamp if available
 5. WHERE no summary exists, THE Backend API SHALL return a preview of the original content (first 200 chars)
 ### Requirement 8: Backward Compatibility
 **User Story:** As a developer, I want the system to work with existing articles, so that no data migration is required.
 #### Acceptance Criteria
 1. THE Crawler Service SHALL work with articles that don't have summaries
 2. THE Backend API SHALL handle articles with or without summaries gracefully
 3. WHERE an article has no summary, THE Backend API SHALL generate a preview from the content field
 4. THE Crawler Service SHALL not re-process articles that already have summaries
 5. THE system SHALL continue to function if Ollama is disabled or unavailable
 ## Non-Functional Requirements
 ### Performance
 - Summarization SHALL complete within 30 seconds per article
 - The crawler SHALL process at least 10 articles per minute (including summarization)
 - Database operations SHALL not be significantly slower with summary storage
 ### Reliability
 - The system SHALL maintain 99% uptime even if Ollama is unavailable
 - Failed summarizations SHALL not prevent article storage
 - The crawler SHALL recover from Ollama errors without manual intervention
 ### Security
 - Ollama API keys SHALL be stored in environment variables, not in code
 - Article content SHALL not be logged to prevent sensitive data exposure
 - API communication with Ollama SHALL support HTTPS
 ### Scalability
 - The system SHALL support multiple Ollama servers for load balancing (future)
 - The crawler SHALL handle articles of any length (up to 50,000 words)
 - The database schema SHALL support future enhancements (tags, categories, etc.)
 ## Dependencies
 - Ollama server must be running and accessible
 - `requests` Python library for HTTP communication
 - Environment variables properly configured
 - MongoDB with sufficient storage for both content and summaries
 ## Assumptions
 - Ollama server is already set up and configured
 - The phi3:latest model (or configured model) supports summarization tasks
 - Network connectivity between crawler and Ollama server is reliable
 - Articles are in English or the configured Ollama model supports the article language
 ## Future Enhancements
 - Support for multiple languages
 - Customizable summary length
 - Sentiment analysis integration
 - Keyword extraction
 - Category classification
 - Batch summarization for improved performance
 - Caching of summaries to avoid re-processing
--- a/.kiro/specs/ai-article-summarization/tasks.md
+++ b/.kiro/specs/ai-article-summarization/tasks.md
@@ -1,92 +0,0 @@
 # Implementation Plan
 - [x] 1. Create Ollama client module
  - Create `news_crawler/ollama_client.py` with OllamaClient class
  - Implement `summarize_article()` method with prompt construction and API call
  - Implement `is_available()` method for health checks
  - Implement `test_connection()` method for diagnostics
  - Add timeout handling (30 seconds)
  - Add error handling for connection, timeout, and invalid responses
  - _Requirements: 1.1, 1.2, 1.3, 1.4, 1.5, 4.1, 4.2, 4.3, 5.2_
 - [x] 2. Create configuration module for crawler
  - Create `news_crawler/config.py` with Config class
  - Load environment variables (OLLAMA_BASE_URL, OLLAMA_MODEL, OLLAMA_ENABLED, OLLAMA_API_KEY, OLLAMA_TIMEOUT)
  - Add validation for required configuration
  - Add default values for optional configuration
  - _Requirements: 2.1, 2.2, 2.3, 2.4_
 - [x] 3. Integrate Ollama client into crawler service
  - Import OllamaClient in `news_crawler/crawler_service.py`
  - Initialize Ollama client at module level using Config
  - Modify `crawl_rss_feed()` to call summarization after content extraction
  - Add conditional logic to skip summarization if OLLAMA_ENABLED is false
  - Add error handling to continue processing if summarization fails
  - Add logging for summarization start, success, and failure
  - Add rate limiting delay after summarization
  - _Requirements: 1.1, 1.2, 1.3, 1.4, 1.5, 2.3, 2.4, 4.1, 4.5, 5.1, 5.3, 6.1, 6.2, 6.3_
 - [x] 4. Update database schema and storage
  - Modify article document structure in `crawl_rss_feed()` to include:
    - `summary` field (AI-generated summary)
    - `summary_word_count` field
    - `summarized_at` field (timestamp)
  - Update MongoDB upsert logic to handle new fields
  - Add check to skip re-summarization if article already has summary
  - _Requirements: 3.1, 3.2, 3.3, 3.4, 8.4_
 - [x] 5. Update backend API to return summaries
  - Modify `backend/routes/news_routes.py` GET /api/news endpoint
  - Add `summary`, `summary_word_count`, `summarized_at` fields to response
  - Add `has_summary` boolean field to indicate if AI summarization was performed
  - Modify GET /api/news/<url> endpoint to include summary fields
  - Add fallback to content preview if no summary exists
  - _Requirements: 7.1, 7.2, 7.3, 7.4, 7.5, 8.1, 8.2, 8.3_
 - [x] 6. Update database schema documentation
  - Update `backend/DATABASE_SCHEMA.md` with new summary fields
  - Add example document showing summary fields
  - Document the summarization workflow
  - _Requirements: 3.1, 3.2, 3.3_
 - [x] 7. Add environment variable configuration
  - Update `backend/env.template` with Ollama configuration
  - Add comments explaining each Ollama setting
  - Document default values
  - _Requirements: 2.1, 2.2_
 - [x] 8. Create test script for Ollama integration
  - Create `news_crawler/test_ollama.py` to test Ollama connection
  - Test summarization with sample article
  - Test error handling (timeout, connection failure)
  - Display configuration and connection status
  - _Requirements: 1.1, 1.2, 1.3, 1.4, 2.1, 2.2, 4.1, 4.2_
 - [x] 9. Update crawler statistics and logging
  - Add summarization statistics to final report in `crawl_all_feeds()`
  - Track total articles summarized vs failed
  - Log average summarization time
  - Display progress indicators during summarization
  - _Requirements: 5.4, 6.1, 6.2, 6.3, 6.4, 6.5_
 - [x] 10. Create documentation for AI summarization
  - Create `news_crawler/AI_SUMMARIZATION.md` explaining the feature
  - Document configuration options
  - Provide troubleshooting guide
  - Add examples of usage
  - _Requirements: 2.1, 2.2, 2.3, 2.4, 6.1, 6.2, 6.3_
 - [x] 11. Update main README with AI summarization info
  - Add section about AI summarization feature
  - Document Ollama setup requirements
  - Add configuration examples
  - Update API endpoint documentation
  - _Requirements: 2.1, 2.2, 7.1, 7.2_
 - [x] 12. Test end-to-end workflow
  - Run crawler with Ollama enabled
  - Verify articles are summarized correctly
  - Check database contains all expected fields
  - Test API endpoints return summaries
  - Verify error handling when Ollama is disabled/unavailable
  - _Requirements: 1.1, 1.2, 1.3, 1.4, 1.5, 3.1, 3.2, 3.3, 3.4, 4.1, 4.2, 4.3, 4.4, 4.5, 7.1, 7.2, 7.3, 7.4, 7.5, 8.1, 8.2, 8.3, 8.4, 8.5_
--- a/.kiro/specs/article-title-translation/design.md
+++ b/.kiro/specs/article-title-translation/design.md
@@ -1,328 +0,0 @@
 # Design Document: Article Title Translation
 ## Overview
 This feature extends the existing Ollama AI integration to translate German article titles to English during the crawling process. The translation will be performed immediately after article content extraction and before AI summarization. Both the original German title and English translation will be stored in the MongoDB article document, and the newsletter template will be updated to display the English title prominently with the original as a subtitle.
 The design leverages the existing Ollama infrastructure (same server, configuration, and error handling patterns) to minimize complexity and maintain consistency with the current summarization feature.
 ## Architecture
 ### Component Interaction Flow
 ```
 RSS Feed Entry
    ↓
 Crawler Service (extract_article_content)
    ↓
 Article Data (with German title)
    ↓
 Ollama Client (translate_title) ← New Method
    ↓
 Translation Result
    ↓
 Crawler Service (prepare article_doc)
    ↓
 MongoDB (articles collection with title + title_en)
    ↓
 Newsletter Service (fetch articles)
    ↓
 Newsletter Template (display English title + German subtitle)
    ↓
 Email to Subscribers
 ```
 ### Integration Points
 1. **Ollama Client** - Add new `translate_title()` method alongside existing `summarize_article()` method
 2. **Crawler Service** - Call translation after content extraction, before summarization
 3. **Article Document Schema** - Add `title_en` and `translated_at` fields
 4. **Newsletter Template** - Update title display logic to show English/German titles
 ## Components and Interfaces
 ### 1. Ollama Client Extension
 **New Method: `translate_title(title, target_language='English')`**
 ```python
 def translate_title(self, title, target_language='English'):
    """
    Translate article title to target language
    Args:
        title (str): Original German title
        target_language (str): Target language (default: 'English')
    Returns:
        dict: {
            'success': bool,
            'translated_title': str or None,
            'error': str or None,
            'duration': float
        }
    """
 ```
 **Implementation Details:**
 - **Prompt Engineering**: Clear, concise prompt instructing the model to translate only the headline without explanations
 - **Temperature**: 0.3 (lower than summarization's 0.7) for more consistent, deterministic translations
 - **Token Limit**: 100 tokens (sufficient for title-length outputs)
 - **Response Cleaning**: 
  - Remove surrounding quotes (single and double)
  - Extract first line only (ignore any extra text)
  - Trim whitespace
 - **Error Handling**: Same pattern as `summarize_article()` - catch timeouts, connection errors, HTTP errors
 - **Validation**: Check for empty title input before making API call
 ### 2. Crawler Service Integration
 **Location**: In `crawl_rss_feed()` function, after content extraction
 **Execution Order**:
 1. Extract article content (existing)
 2. **Translate title** (new)
 3. Summarize article (existing)
 4. Save to database (modified)
 **Implementation Pattern**:
 ```python
 # After article_data extraction
 translation_result = None
 original_title = article_data.get('title') or entry.get('title', '')
 if Config.OLLAMA_ENABLED:
    # Translate title
    print(f"   🌐 Translating title...")
    translation_result = ollama_client.translate_title(original_title)
    if translation_result and translation_result['success']:
        print(f"   ✓ Title translated ({translation_result['duration']:.1f}s)")
    else:
        print(f"   ⚠ Translation failed: {translation_result['error']}")
    # Then summarize (existing code)
    ...
 ```
 **Console Output Format**:
 - Success: `✓ Title translated (0.8s)`
 - Failure: `⚠ Translation failed: Request timed out`
 ### 3. Data Models
 **MongoDB Article Document Schema Extension**:
 ```javascript
 {
  // Existing fields
  title: String,              // Original German title
  author: String,
  link: String,
  content: String,
  summary: String,
  word_count: Number,
  summary_word_count: Number,
  source: String,
  category: String,
  published_at: Date,
  crawled_at: Date,
  summarized_at: Date,
  created_at: Date,
  // New fields
  title_en: String,           // English translation of title (nullable)
  translated_at: Date         // Timestamp when translation completed (nullable)
 }
 ```
 **Field Behavior**:
 - `title_en`: NULL if translation fails or Ollama is disabled
 - `translated_at`: NULL if translation fails, set to `datetime.utcnow()` on success
 ### 4. Newsletter Template Updates
 **Current Title Display**:
 ```html
 <h2 style="...">
    {{ article.title }}
 </h2>
 ```
 **New Title Display Logic**:
 ```html
 <!-- Primary title: English if available, otherwise German -->
 <h2 style="margin: 12px 0 8px 0; font-size: 19px; font-weight: 700; line-height: 1.3; color: #1a1a1a;">
    {{ article.title_en if article.title_en else article.title }}
 </h2>
 <!-- Subtitle: Original German title (only if English translation exists and differs) -->
 {% if article.title_en and article.title_en != article.title %}
 <p style="margin: 0 0 12px 0; font-size: 13px; color: #999999; font-style: italic;">
    Original: {{ article.title }}
 </p>
 {% endif %}
 ```
 **Display Rules**:
 1. If `title_en` exists and differs from `title`: Show English as primary, German as subtitle
 2. If `title_en` is NULL or same as `title`: Show only the original title
 3. Subtitle styling: Smaller font (13px), gray color (#999999), italic
 ## Error Handling
 ### Translation Failure Scenarios
 | Scenario | Behavior | User Impact |
 |----------|----------|-------------|
 | Ollama server unavailable | Skip translation, continue with summarization | Newsletter shows German title only |
 | Translation timeout | Log error, store NULL in title_en | Newsletter shows German title only |
 | Empty title input | Return error immediately, skip API call | Newsletter shows German title only |
 | Ollama disabled in config | Skip translation entirely | Newsletter shows German title only |
 | Network error | Catch exception, log error, continue | Newsletter shows German title only |
 ### Error Handling Principles
 1. **Non-blocking**: Translation failures never prevent article processing
 2. **Graceful degradation**: Fall back to original German title
 3. **Consistent logging**: All errors logged with descriptive messages
 4. **No retry logic**: Single attempt per article (same as summarization)
 5. **Silent failures**: Newsletter displays seamlessly regardless of translation status
 ### Console Output Examples
 **Success Case**:
 ```
 🔍 Crawling: Neuer U-Bahn-Ausbau in München geplant...
   🌐 Translating title...
   ✓ Title translated (0.8s)
   🤖 Summarizing with AI...
   ✓ Summary: 45 words (from 320 words, 2.3s)
   ✓ Saved (320 words)
 ```
 **Translation Failure Case**:
 ```
 🔍 Crawling: Neuer U-Bahn-Ausbau in München geplant...
   🌐 Translating title...
   ⚠ Translation failed: Request timed out after 30 seconds
   🤖 Summarizing with AI...
   ✓ Summary: 45 words (from 320 words, 2.3s)
   ✓ Saved (320 words)
 ```
 ## Testing Strategy
 ### Unit Testing
 **Ollama Client Tests** (`test_ollama_client.py`):
 1. Test successful translation with valid German title
 2. Test empty title input handling
 3. Test timeout handling
 4. Test connection error handling
 5. Test response cleaning (quotes, newlines, whitespace)
 6. Test translation with special characters
 7. Test translation with very long titles
 **Test Data Examples**:
 - Simple: "München plant neue U-Bahn-Linie"
 - With quotes: "\"Historischer Tag\" für München"
 - With special chars: "Oktoberfest 2024: 7,5 Millionen Besucher"
 - Long: "Stadtrat beschließt umfassende Maßnahmen zur Verbesserung der Verkehrsinfrastruktur..."
 ### Integration Testing
 **Crawler Service Tests**:
 1. Test article processing with translation enabled
 2. Test article processing with translation disabled
 3. Test article processing when translation fails
 4. Test database document structure includes new fields
 5. Test console output formatting
 ### Manual Testing
 **End-to-End Workflow**:
 1. Enable Ollama in configuration
 2. Trigger crawl with `max_articles=2`
 3. Verify console shows translation status
 4. Check MongoDB for `title_en` and `translated_at` fields
 5. Send test newsletter
 6. Verify email displays English title with German subtitle
 **Test Scenarios**:
 - Fresh crawl with Ollama enabled
 - Re-crawl existing articles (should skip translation)
 - Crawl with Ollama disabled
 - Crawl with Ollama server stopped (simulate failure)
 ### Performance Testing
 **Metrics to Monitor**:
 - Translation duration per article (target: < 2 seconds)
 - Impact on total crawl time (translation + summarization)
 - Ollama server resource usage
 **Expected Performance**:
 - Translation: ~0.5-1.5 seconds per title
 - Total per article: ~3-5 seconds (translation + summarization)
 - Acceptable for batch processing during scheduled crawls
 ## Configuration
 ### No New Configuration Required
 The translation feature uses existing Ollama configuration:
 ```python
 # From config.py (existing)
 OLLAMA_ENABLED = True/False
 OLLAMA_BASE_URL = "http://ollama:11434"
 OLLAMA_MODEL = "phi3:latest"
 OLLAMA_TIMEOUT = 30
 ```
 **Rationale**: Simplifies deployment and maintains consistency. Translation is automatically enabled/disabled with the existing `OLLAMA_ENABLED` flag.
 ## Deployment Considerations
 ### Docker Container Updates
 **Affected Services**:
 - `crawler` service: Needs rebuild to include new translation code
 - `sender` service: Needs rebuild to include updated newsletter template
 **Deployment Steps**:
 1. Update code in `news_crawler/ollama_client.py`
 2. Update code in `news_crawler/crawler_service.py`
 3. Update template in `news_sender/newsletter_template.html`
 4. Rebuild containers: `docker-compose up -d --build crawler sender`
 5. No database migration needed (new fields are nullable)
 ### Backward Compatibility
 **Existing Articles**: Articles without `title_en` will display German title only (graceful fallback)
 **No Breaking Changes**: Newsletter template handles NULL `title_en` values
 ### Rollback Plan
 If issues arise:
 1. Revert code changes
 2. Rebuild containers
 3. Existing articles with `title_en` will continue to work
 4. New articles will only have German titles
 ## Future Enhancements
 ### Potential Improvements (Out of Scope)
 1. **Batch Translation**: Translate multiple titles in single API call for efficiency
 2. **Translation Caching**: Cache common phrases/words to reduce API calls
 3. **Multi-language Support**: Add configuration for target language selection
 4. **Translation Quality Metrics**: Track and log translation quality scores
 5. **Retry Logic**: Implement retry with exponential backoff for failed translations
 6. **Admin API**: Add endpoint to re-translate existing articles
 These enhancements are not included in the current implementation to maintain simplicity and focus on core functionality.
--- a/.kiro/specs/article-title-translation/requirements.md
+++ b/.kiro/specs/article-title-translation/requirements.md
@@ -1,75 +0,0 @@
 # Requirements Document
 ## Introduction
 This feature adds automatic translation of German article titles to English using the Ollama AI service. The translation will occur during the article crawling process and both the original German title and English translation will be stored in the database. The newsletter will display the English title prominently with the original German title as a subtitle when available.
 ## Glossary
 - **Crawler Service**: The Python service that fetches articles from RSS feeds and processes them
 - **Ollama Client**: The Python client that communicates with the Ollama AI server for text processing
 - **Article Document**: The MongoDB document structure that stores article data
 - **Newsletter Template**: The HTML template used to render the email newsletter sent to subscribers
 - **Translation Result**: The response object returned by the Ollama translation function containing the translated title and metadata
 ## Requirements
 ### Requirement 1
 **User Story:** As a newsletter subscriber, I want to see article titles in English, so that I can quickly understand the content without knowing German
 #### Acceptance Criteria
 1. WHEN the Crawler Service processes an article, THE Ollama Client SHALL translate the German title to English
 2. THE Article Document SHALL store both the original German title and the English translation
 3. THE Newsletter Template SHALL display the English title as the primary heading
 4. WHERE an English translation exists, THE Newsletter Template SHALL display the original German title as a subtitle
 5. IF the translation fails, THEN THE Newsletter Template SHALL display the original German title as the primary heading
 ### Requirement 2
 **User Story:** As a system administrator, I want translation to be integrated with the existing Ollama service, so that I don't need to configure additional services
 #### Acceptance Criteria
 1. THE Ollama Client SHALL provide a translate_title method that accepts a German title and returns an English translation
 2. THE translate_title method SHALL use the same Ollama server configuration as the existing summarization feature
 3. THE translate_title method SHALL use a temperature setting of 0.3 for consistent translations
 4. THE translate_title method SHALL limit the response to 100 tokens maximum for title-length outputs
 5. THE translate_title method SHALL return a Translation Result containing success status, translated title, error message, and duration
 ### Requirement 3
 **User Story:** As a developer, I want translation errors to be handled gracefully, so that article processing continues even when translation fails
 #### Acceptance Criteria
 1. IF the Ollama server is unavailable, THEN THE Crawler Service SHALL continue processing articles without translations
 2. IF a translation request times out, THEN THE Crawler Service SHALL log the error and store the article with only the original title
 3. THE Crawler Service SHALL display translation status in the console output during crawling
 4. THE Article Document SHALL include a translated_at timestamp field when translation succeeds
 5. THE Article Document SHALL store NULL in the title_en field when translation fails
 ### Requirement 4
 **User Story:** As a newsletter subscriber, I want translations to be accurate and natural, so that the English titles read fluently
 #### Acceptance Criteria
 1. THE Ollama Client SHALL provide a clear prompt instructing the model to translate German news headlines to English
 2. THE Ollama Client SHALL instruct the model to provide only the translation without explanations
 3. THE Ollama Client SHALL clean the translation output by removing quotes and extra text
 4. THE Ollama Client SHALL extract only the first line of the translation response
 5. THE Ollama Client SHALL trim whitespace from the translated title
 ### Requirement 5
 **User Story:** As a system operator, I want to see translation performance metrics, so that I can monitor the translation feature effectiveness
 #### Acceptance Criteria
 1. THE Crawler Service SHALL log the translation duration for each article
 2. THE Crawler Service SHALL display a success indicator when translation completes
 3. THE Crawler Service SHALL display an error message when translation fails
 4. THE Translation Result SHALL include the duration in seconds
 5. THE Article Document SHALL store the translated_at timestamp for successful translations
--- a/.kiro/specs/article-title-translation/tasks.md
+++ b/.kiro/specs/article-title-translation/tasks.md
@@ -1,47 +0,0 @@
 # Implementation Plan
 - [x] 1. Add translate_title method to Ollama client
  - Create the `translate_title()` method in `news_crawler/ollama_client.py` that accepts a title string and target language parameter
  - Implement the translation prompt that instructs the model to translate German headlines to English without explanations
  - Configure Ollama API call with temperature=0.3 and num_predict=100 for consistent title-length translations
  - Implement response cleaning logic to remove quotes, extract first line only, and trim whitespace
  - Add error handling for timeout, connection errors, HTTP errors, and empty title input
  - Return a dictionary with success status, translated_title, error message, and duration fields
  - _Requirements: 2.1, 2.2, 2.3, 2.4, 2.5, 4.1, 4.2, 4.3, 4.4, 4.5_
 - [x] 2. Integrate translation into crawler service
  - [x] 2.1 Add translation call in crawl_rss_feed function
    - Locate the article processing section in `news_crawler/crawler_service.py` after content extraction
    - Store the original title from article_data or entry
    - Add conditional check for Config.OLLAMA_ENABLED before calling translation
    - Call `ollama_client.translate_title()` with the original title
    - Store the translation_result for later use in article document
    - _Requirements: 1.1, 2.1_
  - [x] 2.2 Add console logging for translation status
    - Add "🌐 Translating title..." message before translation call
    - Add success message with duration: "✓ Title translated (X.Xs)"
    - Add failure message with error: "⚠ Translation failed: {error}"
    - _Requirements: 5.1, 5.2, 5.3_
  - [x] 2.3 Update article document structure
    - Modify the article_doc dictionary to include `title_en` field with translated title or None
    - Add `translated_at` field set to datetime.utcnow() on success or None on failure
    - Ensure the original `title` field still contains the German title
    - _Requirements: 1.2, 3.5_
 - [x] 3. Update newsletter template for bilingual title display
  - Modify `news_sender/newsletter_template.html` to display English title as primary heading when available
  - Add conditional logic to show original German title as subtitle only when English translation exists and differs
  - Style the subtitle with smaller font (13px), gray color (#999999), and italic formatting
  - Ensure fallback to German title when title_en is NULL or missing
  - _Requirements: 1.3, 1.4, 1.5_
 - [x] 4. Test the translation feature end-to-end
  - Rebuild the crawler Docker container with the new translation code
  - Clear existing articles from the database for clean testing
  - Trigger a test crawl with max_articles=2 to process fresh articles
  - Verify console output shows translation status messages
  - Check MongoDB to confirm title_en and translated_at fields are populated
  - Send a test newsletter email to verify English titles display correctly with German subtitles
  - _Requirements: 1.1, 1.2, 1.3, 1.4, 5.1, 5.2, 5.4, 5.5_
--- a/.kiro/specs/email-tracking/design.md
+++ b/.kiro/specs/email-tracking/design.md
@@ -1,407 +0,0 @@
 # Email Tracking System Design
 ## Overview
 The email tracking system enables Munich News Daily to measure subscriber engagement through email opens and link clicks. The system uses industry-standard techniques (tracking pixels and redirect URLs) while maintaining privacy compliance and performance.
 ## Architecture
 ### High-Level Components
 ```
 ┌─────────────────────────────────────────────────────────────┐
 │                    Newsletter System                         │
 │                                                              │
 │  ┌──────────────┐      ┌──────────────┐                    │
 │  │   Sender     │─────▶│   Tracking   │                    │
 │  │   Service    │      │   Generator  │                    │
 │  └──────────────┘      └──────────────┘                    │
 │         │                      │                            │
 │         │                      ▼                            │
 │         │              ┌──────────────┐                    │
 │         │              │   MongoDB    │                    │
 │         │              │  (tracking)  │                    │
 │         │              └──────────────┘                    │
 │         ▼                                                   │
 │  ┌──────────────┐                                          │
 │  │   Email      │                                          │
 │  │   Client     │                                          │
 │  └──────────────┘                                          │
 └─────────────────────────────────────────────────────────────┘
         │                      ▲
         │                      │
         ▼                      │
 ┌─────────────────────────────────────────────────────────────┐
 │                    Backend API Server                        │
 │                                                              │
 │  ┌──────────────┐      ┌──────────────┐                    │
 │  │   Pixel      │      │   Link       │                    │
 │  │   Endpoint   │      │   Redirect   │                    │
 │  └──────────────┘      └──────────────┘                    │
 │         │                      │                            │
 │         └──────────┬───────────┘                            │
 │                    ▼                                         │
 │            ┌──────────────┐                                 │
 │            │   MongoDB    │                                 │
 │            │  (tracking)  │                                 │
 │            └──────────────┘                                 │
 └─────────────────────────────────────────────────────────────┘
 ```
 ### Technology Stack
 - **Backend**: Flask (Python) - existing backend server
 - **Database**: MongoDB - existing database with new collections
 - **Email**: SMTP (existing sender service)
 - **Tracking**: UUID-based unique identifiers
 - **Image**: 1x1 transparent PNG (base64 encoded)
 ## Components and Interfaces
 ### 1. Tracking ID Generator
 **Purpose**: Generate unique tracking identifiers for emails and links
 **Module**: `backend/services/tracking_service.py`
 **Functions**:
 ```python
 def generate_tracking_id() -> str:
    """Generate a unique tracking ID using UUID4"""
    return str(uuid.uuid4())
 def create_newsletter_tracking(newsletter_id: str, subscriber_email: str) -> dict:
    """Create tracking record for a newsletter send"""
    # Returns tracking document with IDs for pixel and links
 ```
 ### 2. Tracking Pixel Endpoint
 **Purpose**: Serve 1x1 transparent PNG and log email opens
 **Endpoint**: `GET /api/track/pixel/<tracking_id>`
 **Flow**:
 1. Receive request with tracking_id
 2. Look up tracking record in database
 3. Log open event (email, timestamp, user-agent)
 4. Return 1x1 transparent PNG image
 5. Handle multiple opens (update last_opened_at)
 **Response**: 
 - Status: 200 OK
 - Content-Type: image/png
 - Body: 1x1 transparent PNG (43 bytes)
 ### 3. Link Tracking Endpoint
 **Purpose**: Track link clicks and redirect to original URL
 **Endpoint**: `GET /api/track/click/<tracking_id>`
 **Flow**:
 1. Receive request with tracking_id
 2. Look up tracking record and original URL
 3. Log click event (email, article_url, timestamp, user-agent)
 4. Redirect to original article URL (302 redirect)
 5. Handle errors gracefully (redirect to homepage if invalid)
 **Response**:
 - Status: 302 Found
 - Location: Original article URL
 - Performance: < 200ms redirect time
 ### 4. Newsletter Template Modifier
 **Purpose**: Inject tracking pixel and replace article links
 **Module**: `news_sender/tracking_integration.py`
 **Functions**:
 ```python
 def inject_tracking_pixel(html: str, tracking_id: str, api_url: str) -> str:
    """Inject tracking pixel before closing </body> tag"""
    pixel_url = f"{api_url}/api/track/pixel/{tracking_id}"
    pixel_html = f'<img src="{pixel_url}" width="1" height="1" alt="" />'
    return html.replace('</body>', f'{pixel_html}</body>')
 def replace_article_links(html: str, articles: list, tracking_map: dict, api_url: str) -> str:
    """Replace article links with tracking URLs"""
    # For each article link, replace with tracking URL
 ```
 ### 5. Analytics Service
 **Purpose**: Calculate engagement metrics and identify active users
 **Module**: `backend/services/analytics_service.py`
 **Functions**:
 ```python
 def get_open_rate(newsletter_id: str) -> float:
    """Calculate percentage of subscribers who opened"""
 def get_click_rate(article_url: str) -> float:
    """Calculate percentage of subscribers who clicked"""
 def get_subscriber_activity_status(email: str) -> str:
    """Return 'active', 'inactive', or 'dormant'"""
 def update_subscriber_activity_statuses():
    """Batch update all subscriber activity statuses"""
 ```
 ## Data Models
 ### Newsletter Sends Collection (`newsletter_sends`)
 Tracks each newsletter sent to each subscriber.
 ```javascript
 {
  _id: ObjectId,
  newsletter_id: String,           // Unique ID for this newsletter batch (date-based)
  subscriber_email: String,        // Recipient email
  tracking_id: String,             // Unique tracking ID for this send (UUID)
  sent_at: DateTime,               // When email was sent
  opened: Boolean,                 // Whether email was opened
  first_opened_at: DateTime,       // First open timestamp (null if not opened)
  last_opened_at: DateTime,        // Most recent open timestamp
  open_count: Number,              // Number of times opened
  created_at: DateTime             // Record creation time
 }
 ```
 **Indexes**:
 - `tracking_id` (unique) - Fast lookup for pixel requests
 - `newsletter_id` - Analytics queries
 - `subscriber_email` - User activity queries
 - `sent_at` - Time-based queries
 ### Link Clicks Collection (`link_clicks`)
 Tracks individual link clicks.
 ```javascript
 {
  _id: ObjectId,
  tracking_id: String,             // Unique tracking ID for this link (UUID)
  newsletter_id: String,           // Which newsletter this link was in
  subscriber_email: String,        // Who clicked
  article_url: String,             // Original article URL
  article_title: String,           // Article title for reporting
  clicked_at: DateTime,            // When link was clicked
  user_agent: String,              // Browser/client info
  created_at: DateTime             // Record creation time
 }
 ```
 **Indexes**:
 - `tracking_id` (unique) - Fast lookup for redirect requests
 - `newsletter_id` - Analytics queries
 - `article_url` - Article performance queries
 - `subscriber_email` - User activity queries
 ### Subscriber Activity Collection (`subscriber_activity`)
 Aggregated activity status for each subscriber.
 ```javascript
 {
  _id: ObjectId,
  email: String,                   // Subscriber email (unique)
  status: String,                  // 'active', 'inactive', or 'dormant'
  last_opened_at: DateTime,        // Most recent email open
  last_clicked_at: DateTime,       // Most recent link click
  total_opens: Number,             // Lifetime open count
  total_clicks: Number,            // Lifetime click count
  newsletters_received: Number,    // Total newsletters sent
  newsletters_opened: Number,      // Total newsletters opened
  updated_at: DateTime             // Last status update
 }
 ```
 **Indexes**:
 - `email` (unique) - Fast lookup
 - `status` - Filter by activity level
 - `last_opened_at` - Time-based queries
 ## Error Handling
 ### Tracking Pixel Failures
 - **Invalid tracking_id**: Return 1x1 transparent PNG anyway (don't break email rendering)
 - **Database error**: Log error, return pixel (fail silently)
 - **Multiple opens**: Update existing record, don't create duplicate
 ### Link Redirect Failures
 - **Invalid tracking_id**: Redirect to website homepage
 - **Database error**: Log error, redirect to homepage
 - **Missing original URL**: Redirect to homepage
 ### Privacy Compliance
 - **Data retention**: Anonymize tracking data after 90 days
  - Remove email addresses
  - Keep aggregated metrics
 - **Opt-out**: Check subscriber preferences before tracking
 - **GDPR deletion**: Provide endpoint to delete all tracking data for a user
 ## Testing Strategy
 ### Unit Tests
 1. **Tracking ID Generation**
   - Test UUID format
   - Test uniqueness
 2. **Pixel Endpoint**
   - Test valid tracking_id returns PNG
   - Test invalid tracking_id returns PNG
   - Test database logging
 3. **Link Redirect**
   - Test valid tracking_id redirects correctly
   - Test invalid tracking_id redirects to homepage
   - Test click logging
 4. **Analytics Calculations**
   - Test open rate calculation
   - Test click rate calculation
   - Test activity status classification
 ### Integration Tests
 1. **End-to-End Newsletter Flow**
   - Send newsletter with tracking
   - Simulate email open (pixel request)
   - Simulate link click
   - Verify database records
 2. **Privacy Compliance**
   - Test data anonymization
   - Test user data deletion
   - Test opt-out handling
 ### Performance Tests
 1. **Redirect Speed**
   - Measure redirect time (target: < 200ms)
   - Test under load (100 concurrent requests)
 2. **Pixel Serving**
   - Test pixel response time
   - Test caching headers
 ## API Endpoints
 ### Tracking Endpoints
 ```
 GET /api/track/pixel/<tracking_id>
 - Returns: 1x1 transparent PNG
 - Logs: Email open event
 GET /api/track/click/<tracking_id>
 - Returns: 302 redirect to article URL
 - Logs: Link click event
 ```
 ### Analytics Endpoints
 ```
 GET /api/analytics/newsletter/<newsletter_id>
 - Returns: Open rate, click rate, engagement metrics
 GET /api/analytics/article/<article_id>
 - Returns: Click count, click rate for specific article
 GET /api/analytics/subscriber/<email>
 - Returns: Activity status, engagement history
 POST /api/analytics/update-activity
 - Triggers: Batch update of subscriber activity statuses
 - Returns: Update count
 ```
 ### Privacy Endpoints
 ```
 DELETE /api/tracking/subscriber/<email>
 - Deletes: All tracking data for subscriber
 - Returns: Deletion confirmation
 POST /api/tracking/anonymize
 - Triggers: Anonymize tracking data older than 90 days
 - Returns: Anonymization count
 ```
 ## Implementation Phases
 ### Phase 1: Core Tracking (MVP)
 - Tracking ID generation
 - Pixel endpoint
 - Link redirect endpoint
 - Database collections
 - Newsletter template integration
 ### Phase 2: Analytics
 - Open rate calculation
 - Click rate calculation
 - Activity status classification
 - Analytics API endpoints
 ### Phase 3: Privacy & Compliance
 - Data anonymization
 - User data deletion
 - Opt-out handling
 - Privacy notices
 ### Phase 4: Optimization
 - Caching for pixel endpoint
 - Performance monitoring
 - Batch processing for activity updates
 ## Security Considerations
 1. **Rate Limiting**: Prevent abuse of tracking endpoints
 2. **Input Validation**: Validate all tracking_ids (UUID format)
 3. **SQL Injection**: Use parameterized queries (MongoDB safe by default)
 4. **Privacy**: Don't expose subscriber emails in URLs
 5. **HTTPS**: Ensure all tracking URLs use HTTPS in production
 ## Configuration
 Add to `backend/.env`:
 ```env
 # Tracking Configuration
 TRACKING_ENABLED=true
 TRACKING_API_URL=http://localhost:5000
 TRACKING_DATA_RETENTION_DAYS=90
 ```
 ## Monitoring and Metrics
 ### Key Metrics to Track
 1. **Email Opens**
   - Overall open rate
   - Open rate by newsletter
   - Time to first open
 2. **Link Clicks**
   - Overall click rate
   - Click rate by article
   - Click-through rate (CTR)
 3. **Subscriber Engagement**
   - Active subscriber count
   - Inactive subscriber count
   - Dormant subscriber count
 4. **System Performance**
   - Pixel response time
   - Redirect response time
   - Database query performance
--- a/.kiro/specs/email-tracking/requirements.md
+++ b/.kiro/specs/email-tracking/requirements.md
@@ -1,77 +0,0 @@
 # Requirements Document
 ## Introduction
 This document outlines the requirements for implementing email tracking functionality in the Munich News Daily newsletter system. The system will track email opens and link clicks to measure subscriber engagement and identify active users.
 ## Glossary
 - **Newsletter System**: The Munich News Daily email sending service
 - **Tracking Pixel**: A 1x1 transparent image embedded in emails to detect opens
 - **Tracking Link**: A redirecting URL that logs clicks before sending users to the actual destination
 - **Subscriber**: A user who receives the newsletter
 - **Email Open**: When a subscriber's email client loads the tracking pixel
 - **Link Click**: When a subscriber clicks a tracked link in the newsletter
 - **Engagement Metrics**: Data about subscriber interactions with newsletters
 ## Requirements
 ### Requirement 1: Track Email Opens
 **User Story:** As a newsletter administrator, I want to track when subscribers open emails, so that I can measure engagement and identify active users.
 #### Acceptance Criteria
 1. WHEN a newsletter is sent, THE Newsletter System SHALL embed a unique tracking pixel in each email
 2. WHEN a subscriber opens the email, THE Newsletter System SHALL record the open event with timestamp
 3. THE Newsletter System SHALL store the subscriber email, newsletter ID, and open timestamp in the database
 4. THE Newsletter System SHALL serve the tracking pixel as a 1x1 transparent PNG image
 5. THE Newsletter System SHALL handle multiple opens from the same subscriber for the same newsletter
 ### Requirement 2: Track Link Clicks
 **User Story:** As a newsletter administrator, I want to track when subscribers click on article links, so that I can understand which content is most engaging.
 #### Acceptance Criteria
 1. WHEN a newsletter is generated, THE Newsletter System SHALL replace all article links with unique tracking URLs
 2. WHEN a subscriber clicks a tracking URL, THE Newsletter System SHALL record the click event with timestamp
 3. WHEN a click is recorded, THE Newsletter System SHALL redirect the subscriber to the original article URL
 4. THE Newsletter System SHALL store the subscriber email, article link, and click timestamp in the database
 5. THE Newsletter System SHALL complete the redirect within 200 milliseconds
 ### Requirement 3: Generate Engagement Reports
 **User Story:** As a newsletter administrator, I want to view engagement metrics, so that I can understand subscriber behavior and content performance.
 #### Acceptance Criteria
 1. THE Newsletter System SHALL provide an API endpoint to retrieve open rates by newsletter
 2. THE Newsletter System SHALL provide an API endpoint to retrieve click rates by article
 3. THE Newsletter System SHALL calculate the percentage of subscribers who opened each newsletter
 4. THE Newsletter System SHALL calculate the percentage of subscribers who clicked each article link
 5. THE Newsletter System SHALL identify subscribers who have not opened emails in the last 30 days
 ### Requirement 4: Privacy and Compliance
 **User Story:** As a newsletter administrator, I want to respect subscriber privacy, so that the system complies with privacy regulations.
 #### Acceptance Criteria
 1. THE Newsletter System SHALL include a privacy notice in the newsletter footer explaining tracking
 2. THE Newsletter System SHALL anonymize tracking data after 90 days by removing email addresses
 3. THE Newsletter System SHALL provide an API endpoint to delete all tracking data for a specific subscriber
 4. THE Newsletter System SHALL not track subscribers who have opted out of tracking
 5. WHERE a subscriber unsubscribes, THE Newsletter System SHALL delete all their tracking data
 ### Requirement 5: Identify Active Users
 **User Story:** As a newsletter administrator, I want to identify active subscribers, so that I can segment my audience and improve targeting.
 #### Acceptance Criteria
 1. THE Newsletter System SHALL classify a subscriber as "active" if they opened an email in the last 30 days
 2. THE Newsletter System SHALL classify a subscriber as "inactive" if they have not opened an email in 30-60 days
 3. THE Newsletter System SHALL classify a subscriber as "dormant" if they have not opened an email in over 60 days
 4. THE Newsletter System SHALL provide an API endpoint to retrieve subscriber activity status
 5. THE Newsletter System SHALL update subscriber activity status daily
--- a/.kiro/specs/email-tracking/tasks.md
+++ b/.kiro/specs/email-tracking/tasks.md
@@ -1,170 +0,0 @@
 # Implementation Plan
 ## Phase 1: Core Tracking Infrastructure
 - [x] 1. Set up database collections and indexes
  - Create MongoDB collections: `newsletter_sends`, `link_clicks`, `subscriber_activity`
  - Add indexes for performance: `tracking_id` (unique), `newsletter_id`, `subscriber_email`, `sent_at`
  - Write database initialization script
  - _Requirements: 1.3, 2.4_
 - [x] 2. Implement tracking service
 - [x] 2.1 Create tracking ID generator
  - Write `generate_tracking_id()` function using UUID4
  - Write `create_newsletter_tracking()` function to create tracking records
  - Add configuration for tracking enable/disable
  - _Requirements: 1.1, 2.1_
 - [x] 2.2 Implement tracking pixel endpoint
  - Create Flask route `GET /api/track/pixel/<tracking_id>`
  - Generate 1x1 transparent PNG (base64 encoded)
  - Log email open event to `newsletter_sends` collection
  - Handle multiple opens (update `last_opened_at` and `open_count`)
  - Return PNG with proper headers (Content-Type: image/png)
  - _Requirements: 1.2, 1.3, 1.4, 1.5_
 - [x] 2.3 Implement link redirect endpoint
  - Create Flask route `GET /api/track/click/<tracking_id>`
  - Look up original article URL from tracking record
  - Log click event to `link_clicks` collection
  - Redirect to original URL with 302 status
  - Handle invalid tracking_id (redirect to homepage)
  - Ensure redirect completes within 200ms
  - _Requirements: 2.2, 2.3, 2.4, 2.5_
 - [x] 2.4 Write unit tests for tracking endpoints
  - Test pixel endpoint returns PNG for valid tracking_id
  - Test pixel endpoint returns PNG for invalid tracking_id (fail silently)
  - Test link redirect works correctly
  - Test link redirect handles invalid tracking_id
  - Test database logging for opens and clicks
  - _Requirements: 1.2, 1.4, 2.2, 2.3_
 ## Phase 2: Newsletter Integration
 - [x] 3. Integrate tracking into sender service
 - [x] 3.1 Create tracking integration module
  - Write `inject_tracking_pixel()` function to add pixel to HTML
  - Write `replace_article_links()` function to replace links with tracking URLs
  - Write `generate_tracking_urls()` function to create tracking records for all links
  - Add tracking configuration to sender service
  - _Requirements: 1.1, 2.1_
 - [x] 3.2 Modify newsletter sending flow
  - Update `send_newsletter()` to generate tracking IDs for each subscriber
  - Create tracking records in database before sending
  - Inject tracking pixel into newsletter HTML
  - Replace article links with tracking URLs
  - Store newsletter_id and tracking metadata
  - _Requirements: 1.1, 1.3, 2.1, 2.4_
 - [x] 3.3 Update newsletter template
  - Ensure template supports dynamic tracking pixel injection
  - Ensure article links are properly structured for replacement
  - Add privacy notice to footer about tracking
  - _Requirements: 4.1_
 - [x] 3.4 Test newsletter with tracking
  - Send test newsletter with tracking enabled
  - Verify tracking pixel is embedded correctly
  - Verify article links are replaced with tracking URLs
  - Test email open tracking works
  - Test link click tracking works
  - _Requirements: 1.1, 1.2, 2.1, 2.2_
 ## Phase 3: Analytics and Reporting
 - [x] 4. Implement analytics service
 - [x] 4.1 Create analytics calculation functions
  - Write `get_open_rate(newsletter_id)` function
  - Write `get_click_rate(article_url)` function
  - Write `get_newsletter_metrics(newsletter_id)` function for overall stats
  - Write `get_article_performance(article_url)` function
  - _Requirements: 3.1, 3.2, 3.3, 3.4_
 - [x] 4.2 Implement subscriber activity classification
  - Write `get_subscriber_activity_status(email)` function
  - Classify as 'active' (opened in last 30 days)
  - Classify as 'inactive' (no opens in 30-60 days)
  - Classify as 'dormant' (no opens in 60+ days)
  - Write `update_subscriber_activity_statuses()` batch function
  - _Requirements: 5.1, 5.2, 5.3, 5.4, 5.5_
 - [x] 4.3 Create analytics API endpoints
  - Create `GET /api/analytics/newsletter/<newsletter_id>` endpoint
  - Create `GET /api/analytics/article/<article_id>` endpoint
  - Create `GET /api/analytics/subscriber/<email>` endpoint
  - Create `POST /api/analytics/update-activity` endpoint
  - Return JSON with engagement metrics
  - _Requirements: 3.1, 3.2, 3.4, 5.4_
 - [x] 4.4 Write unit tests for analytics
  - Test open rate calculation
  - Test click rate calculation
  - Test activity status classification
  - Test edge cases (no opens, no clicks)
  - _Requirements: 3.3, 3.4, 5.1, 5.2, 5.3_
 ## Phase 4: Privacy and Compliance
 - [x] 5. Implement privacy features
 - [x] 5.1 Create data anonymization function
  - Write function to anonymize tracking data older than 90 days
  - Remove email addresses from old records
  - Keep aggregated metrics
  - Create scheduled task to run daily
  - _Requirements: 4.2_
 - [x] 5.2 Implement user data deletion
  - Create `DELETE /api/tracking/subscriber/<email>` endpoint
  - Delete all tracking records for subscriber
  - Delete from `newsletter_sends`, `link_clicks`, `subscriber_activity`
  - Return confirmation response
  - _Requirements: 4.3, 4.5_
 - [x] 5.3 Add tracking opt-out support
  - Add `tracking_enabled` field to subscribers collection
  - Check opt-out status before creating tracking records
  - Skip tracking for opted-out subscribers
  - Update newsletter sending to respect opt-out
  - _Requirements: 4.4_
 - [x] 5.4 Create anonymization endpoint
  - Create `POST /api/tracking/anonymize` endpoint
  - Trigger anonymization of old data
  - Return count of anonymized records
  - Add authentication/authorization
  - _Requirements: 4.2_
 - [x] 5.5 Write privacy compliance tests
  - Test data anonymization works correctly
  - Test user data deletion removes all records
  - Test opt-out prevents tracking
  - Test anonymization preserves aggregated metrics
  - _Requirements: 4.2, 4.3, 4.4, 4.5_
 ## Phase 5: Configuration and Documentation
 - [x] 6. Add configuration and environment setup
  - Add `TRACKING_ENABLED` to environment variables
  - Add `TRACKING_API_URL` configuration
  - Add `TRACKING_DATA_RETENTION_DAYS` configuration
  - Update `.env.template` with tracking variables
  - Update configuration documentation
  - _Requirements: All_
 - [x] 7. Update database schema documentation
  - Document `newsletter_sends` collection schema
  - Document `link_clicks` collection schema
  - Document `subscriber_activity` collection schema
  - Add indexes documentation
  - Update DATABASE_SCHEMA.md
  - _Requirements: 1.3, 2.4, 5.4_
 - [x] 8. Create tracking usage documentation
  - Document how to enable/disable tracking
  - Document analytics API endpoints
  - Document privacy features
  - Add examples of querying tracking data
  - Create README for tracking system
  - _Requirements: All_
--- a/25
+++ b/25
@@ -0,0 +1,25 @@
 pipeline {
    agent any
    stages {
        stage('Security Scan') {
            steps {
                withCredentials([string(credentialsId: 'nvd-api-key', variable: 'NVD_API_KEY')]) {
                    // Run OWASP Dependency Check using the specific installation configured in Jenkins
                    // Using NVD API Key to avoid rate limiting
                    dependencyCheck additionalArguments: "--scan ./ --format ALL --nvdApiKey ${NVD_API_KEY}", odcInstallation: 'depcheck'
                }
            }
        }
    }
    post {
        always {
            // Publish the results
            dependencyCheckPublisher pattern: 'dependency-check-report.xml'
            // Archive the reports
            archiveArtifacts allowEmptyArchive: true, artifacts: 'dependency-check-report.html'
        }
    }
 }
--- a/QUICKSTART.md
+++ b/QUICKSTART.md
@@ -1,56 +1,36 @@
-# Quick Start Guide
+# ⚡ Quick Start Guide
 Get Munich News Daily running in 5 minutes!
-## Prerequisites
+## 📋 Prerequisites
 - **Docker** & **Docker Compose** installed
 - **4GB+ RAM** (for AI models)
 - *(Optional)* NVIDIA GPU for faster processing
- Docker & Docker Compose installed
+## 🚀 Setup Steps
 - 4GB+ RAM (for Ollama AI models)
 - (Optional) NVIDIA GPU for 5-10x faster AI processing
 ## Setup
 ### 1. Configure Environment
 ```bash
 # Copy example environment file
 cp backend/.env.example backend/.env
 # Edit with your settings (required: email configuration)
 nano backend/.env
 ```
 **Required:** Update `SMTP_SERVER`, `EMAIL_USER`, and `EMAIL_PASSWORD`.
-**Minimum required settings:**
+### 2. Start the System
 ```env
 SMTP_SERVER=smtp.gmail.com
 SMTP_PORT=587
 EMAIL_USER=your-email@gmail.com
 EMAIL_PASSWORD=your-app-password
 ```
 ### 2. Start System
 ```bash
-# Option 1: Auto-detect GPU and start (recommended)
+# Auto-detects GPU capabilities and starts services
 ./start-with-gpu.sh
-# Option 2: Start without GPU
+# Watch installation progress (first time model download ~2GB)
 docker-compose up -d
 # View logs
 docker-compose logs -f
 # Wait for Ollama model download (first time only, ~2-5 minutes)
 docker-compose logs -f ollama-setup
 ```
-**Note:** First startup downloads the phi3:latest AI model (2.2GB). This happens automatically.
+### 3. Add News Sources
 ### 3. Add RSS Feeds
 ```bash
-mongosh munich_news
+# Connect to database
 docker-compose exec mongodb mongosh munich_news
 # Paste this into the mongo shell:
 db.rss_feeds.insertMany([
  {
    name: "Süddeutsche Zeitung München",
@@ -65,11 +45,9 @@ db.rss_feeds.insertMany([
 ])
 ```
-### 4. Add Subscribers
+### 4. Add Yourself as Subscriber
 ```bash
-mongosh munich_news
+# Still in mongo shell:
 db.subscribers.insertOne({
  email: "your-email@example.com",
  active: true,
@@ -78,90 +56,35 @@ db.subscribers.insertOne({
 })
 ```
-### 5. Test It
+### 5. Verify Installation
 ```bash
-# Test crawler
+# 1. Run the crawler manually to fetch news
 docker-compose exec crawler python crawler_service.py 5
-# Test newsletter
+# 2. Send a test email to yourself
 docker-compose exec sender python sender_service.py test your-email@example.com
 ```
-## What Happens Next?
+## 🎮 Dashboard Access
-The system will automatically:
+Once running, access the services:
- **Backend API**: Runs continuously at http://localhost:5001 for tracking and analytics
+- **Dashboard**: [http://localhost:3000](http://localhost:3000)
- **6:00 AM Berlin time**: Crawl news articles
+- **API**: [http://localhost:5001](http://localhost:5001)
 - **7:00 AM Berlin time**: Send newsletter to subscribers
-## View Results
+## ⏭️ What's Next?
 The system is now fully automated:
 1.  **6:00 AM**: Crawls news and generates AI summaries.
 2.  **7:00 AM**: Sends the daily newsletter.
 ### Useful Commands
 ```bash
-# Check articles
+# Stop everything
 mongosh munich_news
 db.articles.find().sort({ crawled_at: -1 }).limit(5)
 # Check logs
 docker-compose logs -f crawler
 docker-compose logs -f sender
 ```
 ## Common Commands
 ```bash
 # Stop system
 docker-compose down
-# Restart system
+# View logs for a service
-docker-compose restart
+docker-compose logs -f crawler
-# View logs
+# Update code & rebuild
 docker-compose logs -f
 # Rebuild after changes
 docker-compose up -d --build
 ```
 ## New Features
 ### GPU Acceleration (5-10x Faster)
 Enable GPU support for faster AI processing:
 ```bash
 ./check-gpu.sh          # Check if GPU is available
 ./start-with-gpu.sh     # Start with GPU support
 ```
 See [docs/GPU_SETUP.md](docs/GPU_SETUP.md) for details.
 ### Send Newsletter to All Subscribers
 ```bash
 # Send newsletter to all active subscribers
 curl -X POST http://localhost:5001/api/admin/send-newsletter \
  -H "Content-Type: application/json" \
  -d '{"max_articles": 10}'
 ```
 ### Security Features
 - ✅ Only Backend API exposed (port 5001)
 - ✅ MongoDB internal-only (secure)
 - ✅ Ollama internal-only (secure)
 - ✅ All services communicate via internal Docker network
 ## Need Help?
 - **Documentation Index**: [docs/INDEX.md](docs/INDEX.md)
 - **GPU Setup**: [docs/GPU_SETUP.md](docs/GPU_SETUP.md)
 - **API Reference**: [docs/ADMIN_API.md](docs/ADMIN_API.md)
 - **Security Guide**: [docs/SECURITY_NOTES.md](docs/SECURITY_NOTES.md)
 - **Full Documentation**: [README.md](README.md)
 ## Next Steps
 1. ✅ **Enable GPU acceleration** - [docs/GPU_SETUP.md](docs/GPU_SETUP.md)
 2. Set up tracking API (optional)
 3. Customize newsletter template
 4. Add more RSS feeds
 5. Monitor engagement metrics
 6. Review security settings - [docs/SECURITY_NOTES.md](docs/SECURITY_NOTES.md)
 That's it! Your automated news system is running. 🎉
--- a/README.md
+++ b/README.md
@@ -1,454 +1,193 @@
 # Munich News Daily - Automated Newsletter System
-A fully automated news aggregation and newsletter system that crawls Munich news sources, generates AI summaries, and sends daily newsletters with engagement tracking.
+A fully automated news aggregation system that crawls Munich news sources, generates AI-powered summaries, tracks local transport disruptions, and delivers personalized daily newsletters.
 ![Munich News Daily](https://via.placeholder.com/800x400?text=Munich+News+Daily+Dashboard)
 ## ✨ Key Features
- **🤖 AI-Powered Clustering** - Automatically detects duplicate stories from different sources
+- **🤖 AI-Powered Clustering** - Smartly detects duplicate stories and groups related articles using ChromaDB vector search.
- **📰 Neutral Summaries** - Combines multiple perspectives into balanced coverage
+- **📝 Neutral Summaries** - Generates balanced, multi-perspective summaries using local LLMs (Ollama).
- **🎯 Smart Prioritization** - Shows most important stories first (multi-source coverage)
+- **🚇 Transport Updates** - Real-time tracking of Munich public transport (MVG) disruptions options.
- **📊 Engagement Tracking** - Open rates, click tracking, and analytics
+- **🎯 Smart Prioritization** - Ranks stories based on relevance and user preferences.
- **⚡ GPU Acceleration** - 5-10x faster AI processing with GPU support
+- **🎨 Personalized Newsletters** - diverse content delivery system.
- **🔒 GDPR Compliant** - Privacy-first with data retention controls
+- **📊 Engagement Analytics** - Detailed tracking of open rates, click-throughs, and user interests.
-
+- **⚡ GPU Acceleration** - Integrated support for NVIDIA GPUs for faster AI processing.
-**🚀 NEW:** GPU acceleration support for 5-10x faster AI processing! See [docs/GPU_SETUP.md](docs/GPU_SETUP.md)
+- **🔒 Privacy First** - GDPR-compliant with automatic data retention policies and anonymization.
 ## 🚀 Quick Start
 For a detailed 5-minute setup guide, see [QUICKSTART.md](QUICKSTART.md).
 ```bash
 # 1. Configure environment
 cp backend/.env.example backend/.env
 # Edit backend/.env with your email settings
-# 2. Start everything
+# 2. Start everything (Auto-detects GPU)
-docker-compose up -d
+./start-with-gpu.sh
-# 3. View logs
+# Questions?
-docker-compose logs -f
+# See logs: docker-compose logs -f
 ```
-That's it! The system will automatically:
+The system will automatically:
- **Frontend**: Web interface and admin dashboard (http://localhost:3000)
+1.  **6:00 AM**: Crawl news & transport updates.
- **Backend API**: Runs continuously for tracking and analytics (http://localhost:5001)
+2.  **6:30 AM**: Generate AI summaries & clusters.
- **6:00 AM Berlin time**: Crawl news articles and generate summaries
+3.  **7:00 AM**: Send personalized newsletters.
 - **7:00 AM Berlin time**: Send newsletter to all subscribers
-### Access Points
+## 📋 System Architecture
- **Newsletter Page**: http://localhost:3000
+The system is built as a set of microservices orchestrated by Docker Compose.
 - **Admin Dashboard**: http://localhost:3000/admin.html
 - **Backend API**: http://localhost:5001
-📖 **New to the project?** See [QUICKSTART.md](QUICKSTART.md) for a detailed 5-minute setup guide.
+```mermaid
 graph TD
    User[Subscribers] -->|Email| Sender[Newsletter Sender]
    User -->|Web| Frontend[React Frontend]
    Frontend -->|API| Backend[Backend API]
-🚀 **GPU Acceleration:** Enable 5-10x faster AI processing with [GPU Setup Guide](docs/GPU_SETUP.md)
+    subgraph "Core Services"
        Crawler[News Crawler]
        Transport[Transport Crawler]
        Sender
        Backend
    end
-## 📋 System Overview
+    subgraph "Data & AI"
        Mongo[(MongoDB)]
        Redis[(Redis)]
        Chroma[(ChromaDB)]
        Ollama[Ollama AI]
    end
-```
+    Crawler -->|Save| Mongo
-6:00 AM → News Crawler
+    Crawler -->|Embeddings| Chroma
-          ↓
+    Crawler -->|Summarize| Ollama
-          Fetches articles from RSS feeds
+    
-          Extracts full content
+    Transport -->|Save| Mongo
-          Generates AI summaries
+    
-          Saves to MongoDB
+    Sender -->|Read| Mongo
-          ↓
+    Sender -->|Track| Backend
-7:00 AM → Newsletter Sender
+    
-          ↓
+    Backend -->|Read/Write| Mongo
-          Waits for crawler to finish
+    Backend -->|Cache| Redis
          Fetches today's articles
          Generates newsletter with tracking
          Sends to all subscribers
          ↓
          ✅ Done! Repeat tomorrow
 ```
-## 🏗️ Architecture
+### Core Components
-### Components
+| Service | Description | Port |
 |---------|-------------|------|
 | **Frontend** | React-based user dashboard and admin interface. | 3000 |
 | **Backend API** | Flask API for tracking, analytics, and management. | 5001 |
 | **News Crawler** | Fetches RSS feeds, extracts content, and runs AI clustering. | - |
 | **Transport Crawler** | Monitors MVG (Munich Transport) for delays and disruptions. | - |
 | **Newsletter Sender** | Manages subscribers, generates templates, and sends emails. | - |
 | **Ollama** | Local LLM runner for on-premise AI (Phi-3, Llama3, etc.). | - |
 | **ChromaDB** | Vector database for semantic search and article clustering. | - |
- **Ollama**: AI service for summarization and translation (internal only, GPU-accelerated)
+## 📂 Project Structure
 - **MongoDB**: Data storage (articles, subscribers, tracking) (internal only)
 - **Backend API**: Flask API for tracking and analytics (port 5001 - only exposed service)
 - **News Crawler**: Automated RSS feed crawler with AI summarization (internal only)
 - **Newsletter Sender**: Automated email sender with tracking (internal only)
 - **Frontend**: React dashboard (optional)
-### Technology Stack
+```text
 munich-news/
 ├── backend/            # Flask API for tracking & analytics
 ├── frontend/           # React dashboard & admin UI
 ├── news_crawler/       # RSS fetcher & AI summarizer service
 ├── news_sender/        # Email generation & dispatch service
 ├── transport_crawler/  # MVG transport disruption monitor
 ├── docker-compose.yml  # Main service orchestration
 └── docs/               # Detailed documentation
 ```
- Python 3.11
+## 🛠️ Installation & Setup
 - MongoDB 7.0
 - Ollama (phi3:latest model for AI)
 - Docker & Docker Compose
 - Flask (API)
 - Schedule (automation)
 - Jinja2 (email templates)
 ## 📦 Installation
 ### Prerequisites
 - Docker & Docker Compose
 - 4GB+ RAM (for Ollama AI models)
 - (Optional) NVIDIA GPU for 5-10x faster AI processing
 ### Setup
 1.  **Clone the repository**
    ```bash
-   git clone <repository-url>
+    git clone https://github.com/yourusername/munich-news.git
    cd munich-news
    ```
-2. **Configure environment**
+2.  **Environment Configuration**
    ```bash
    cp backend/.env.example backend/.env
-   # Edit backend/.env with your settings
+    nano backend/.env
    ```
    *Critical settings:* `SMTP_SERVER`, `EMAIL_USER`, `EMAIL_PASSWORD`.
-3. **Configure Ollama (AI features)**
+3.  **Start the System**
    ```bash
-   # Option 1: Use integrated Docker Compose Ollama (recommended)
+    # Recommended: Helper script (handles GPU & Model setup)
   ./configure-ollama.sh
   # Select option 1
   # Option 2: Use external Ollama server
   # Install from https://ollama.ai/download
   # Then run: ollama pull phi3:latest
   ```
 4. **Start the system**
   ```bash
   # Auto-detect GPU and start (recommended)
    ./start-with-gpu.sh
-   # Or start manually
+    # Alternative: Standard Docker Compose
    docker-compose up -d
   # First time: Wait for Ollama model download (2-5 minutes)
   docker-compose logs -f ollama-setup
    ```
-📖 **For detailed Ollama setup & GPU acceleration:** See [docs/OLLAMA_SETUP.md](docs/OLLAMA_SETUP.md)
+4.  **Initial Setup (First Run)**
-
+    *   The system needs to download the AI model (approx. 2GB).
-💡 **To change AI model:** Edit `OLLAMA_MODEL` in `.env`, then run `./pull-ollama-model.sh`. See [docs/CHANGING_AI_MODEL.md](docs/CHANGING_AI_MODEL.md)
+    *   Watch progress: `docker-compose logs -f ollama-setup`
 ## ⚙️ Configuration
-Edit `backend/.env`:
+Key configuration options in `backend/.env`:
-```env
+| Category | Variable | Description |
-# MongoDB
+|----------|----------|-------------|
-MONGODB_URI=mongodb://localhost:27017/
+| **Email** | `SMTP_SERVER` | SMTP Server (e.g., smtp.gmail.com) |
 | | `EMAIL_USER` | Your sending email address |
 | **AI** | `OLLAMA_MODEL` | Model to use (default: phi3:latest) |
 | **Schedule** | `CRAWLER_TIME` | Time to start crawling (e.g., "06:00") |
 | | `SENDER_TIME` | Time to send emails (e.g., "07:00") |
-# Email (SMTP)
+## 📊 Usage & Monitoring
 SMTP_SERVER=smtp.gmail.com
 SMTP_PORT=587
 EMAIL_USER=your-email@gmail.com
 EMAIL_PASSWORD=your-app-password
-# Newsletter
+### Access Points
-NEWSLETTER_MAX_ARTICLES=10
+*   **Web Dashboard**: [http://localhost:3000](http://localhost:3000) (or configured domain)
-NEWSLETTER_HOURS_LOOKBACK=24
+*   **API**: [http://localhost:5001](http://localhost:5001)
-# Tracking
+### Useful Commands
 TRACKING_ENABLED=true
 TRACKING_API_URL=http://localhost:5001
 TRACKING_DATA_RETENTION_DAYS=90
-# Ollama (AI Summarization)
+**View Logs**
-OLLAMA_ENABLED=true
+```bash
-OLLAMA_BASE_URL=http://127.0.0.1:11434
+docker-compose logs -f [service_name]
-OLLAMA_MODEL=phi3:latest
+# e.g., docker-compose logs -f crawler
 ```
-## 📊 Usage
+**Manual Trigger**
 ### View Logs
 ```bash
-# All services
+# Run News Crawler immediately
 docker-compose logs -f
 # Specific service
 docker-compose logs -f crawler
 docker-compose logs -f sender
 docker-compose logs -f mongodb
 ```
 ### Manual Operations
 ```bash
 # Run crawler manually
 docker-compose exec crawler python crawler_service.py 10
-# Send test newsletter
+# Run Transport Crawler immediately
-docker-compose exec sender python sender_service.py test your-email@example.com
+docker-compose exec transport-crawler python transport_service.py
-# Preview newsletter
+# Send Test Newsletter
-docker-compose exec sender python sender_service.py preview
+docker-compose exec sender python sender_service.py test user@example.com
 ```
-### Database Access
+**Database Access**
 ```bash
 # Connect to MongoDB
 docker-compose exec mongodb mongosh munich_news
 # View articles
 db.articles.find().sort({ crawled_at: -1 }).limit(5).pretty()
 # View subscribers
 db.subscribers.find({ active: true }).pretty()
 # View tracking data
 db.newsletter_sends.find().sort({ created_at: -1 }).limit(10).pretty()
 ```
-## 🔧 Management
+## 🌐 Production Deployment (Traefik)
-### Add RSS Feeds
+This project is configured to work with **Traefik** as a reverse proxy.
 The `docker-compose.yml` includes labels for:
 - `news.dongho.kim` (Frontend)
 - `news-api.dongho.kim` (Backend)
-```bash
+To use this locally, add these to your `/etc/hosts`:
-mongosh munich_news
+```text
-
+127.0.0.1 news.dongho.kim news-api.dongho.kim
 db.rss_feeds.insertOne({
  name: "Source Name",
  url: "https://example.com/rss",
  active: true
 })
 ```
-### Add Subscribers
+For production, ensure your Traefik proxy network is named `proxy` or update the `docker-compose.yml` accordingly.
 ```bash
 mongosh munich_news
 db.subscribers.insertOne({
  email: "user@example.com",
  active: true,
  tracking_enabled: true,
  subscribed_at: new Date()
 })
 ```
 ### View Analytics
 ```bash
 # Newsletter metrics
 curl http://localhost:5001/api/analytics/newsletter/2024-01-15
 # Article performance
 curl http://localhost:5001/api/analytics/article/https://example.com/article
 # Subscriber activity
 curl http://localhost:5001/api/analytics/subscriber/user@example.com
 ```
 ## ⏰ Schedule Configuration
 ### Change Crawler Time (default: 6:00 AM)
 Edit `news_crawler/scheduled_crawler.py`:
 ```python
 schedule.every().day.at("06:00").do(run_crawler)  # Change time
 ```
 ### Change Sender Time (default: 7:00 AM)
 Edit `news_sender/scheduled_sender.py`:
 ```python
 schedule.every().day.at("07:00").do(run_sender)  # Change time
 ```
 After changes:
 ```bash
 docker-compose up -d --build
 ```
 ## 📈 Monitoring
 ### Container Status
 ```bash
 docker-compose ps
 ```
 ### Check Next Scheduled Runs
 ```bash
 # Crawler
 docker-compose logs crawler | grep "Next scheduled run"
 # Sender
 docker-compose logs sender | grep "Next scheduled run"
 ```
 ### Engagement Metrics
 ```bash
 mongosh munich_news
 // Open rate
 var sent = db.newsletter_sends.countDocuments({ newsletter_id: "2024-01-15" })
 var opened = db.newsletter_sends.countDocuments({ newsletter_id: "2024-01-15", opened: true })
 print("Open Rate: " + ((opened / sent) * 100).toFixed(2) + "%")
 // Click rate
 var clicks = db.link_clicks.countDocuments({ newsletter_id: "2024-01-15" })
 print("Click Rate: " + ((clicks / sent) * 100).toFixed(2) + "%")
 ```
 ## 🐛 Troubleshooting
 ### Crawler Not Finding Articles
 ```bash
 # Check RSS feeds
 mongosh munich_news --eval "db.rss_feeds.find({ active: true })"
 # Test manually
 docker-compose exec crawler python crawler_service.py 5
 ```
 ### Newsletter Not Sending
 ```bash
 # Check email config
 docker-compose exec sender python -c "from sender_service import Config; print(Config.SMTP_SERVER)"
 # Test email
 docker-compose exec sender python sender_service.py test your-email@example.com
 ```
 ### Containers Not Starting
 ```bash
 # Check logs
 docker-compose logs
 # Rebuild
 docker-compose up -d --build
 # Reset everything
 docker-compose down -v
 docker-compose up -d
 ```
 ## 🔐 Privacy & Compliance
 ### GDPR Features
 - **Data Retention**: Automatic anonymization after 90 days
 - **Opt-Out**: Subscribers can disable tracking
 - **Data Deletion**: Full data removal on request
 - **Transparency**: Privacy notice in all emails
 ### Privacy Endpoints
 ```bash
 # Delete subscriber data
 curl -X DELETE http://localhost:5001/api/tracking/subscriber/user@example.com
 # Anonymize old data
 curl -X POST http://localhost:5001/api/tracking/anonymize
 # Opt out of tracking
 curl -X POST http://localhost:5001/api/tracking/subscriber/user@example.com/opt-out
 ```
 ## 📚 Documentation
 ### Getting Started
 - **[QUICKSTART.md](QUICKSTART.md)** - 5-minute setup guide
 - **[CONTRIBUTING.md](CONTRIBUTING.md)** - Contribution guidelines
 ### Core Features
 - **[docs/AI_NEWS_AGGREGATION.md](docs/AI_NEWS_AGGREGATION.md)** - AI-powered clustering & neutral summaries
 - **[docs/FEATURES.md](docs/FEATURES.md)** - Complete feature list
 - **[docs/API.md](docs/API.md)** - API endpoints reference
 ### Technical Documentation
 - **[docs/ARCHITECTURE.md](docs/ARCHITECTURE.md)** - System architecture
 - **[docs/SETUP.md](docs/SETUP.md)** - Detailed setup guide
 - **[docs/OLLAMA_SETUP.md](docs/OLLAMA_SETUP.md)** - AI/Ollama configuration
 - **[docs/GPU_SETUP.md](docs/GPU_SETUP.md)** - GPU acceleration setup
 - **[docs/DEPLOYMENT.md](docs/DEPLOYMENT.md)** - Production deployment
 - **[docs/SECURITY.md](docs/SECURITY.md)** - Security best practices
 - **[docs/REFERENCE.md](docs/REFERENCE.md)** - Complete reference
 - **[docs/DEPLOYMENT.md](docs/DEPLOYMENT.md)** - Deployment guide
 - **[docs/API.md](docs/API.md)** - API reference
 - **[docs/DATABASE_SCHEMA.md](docs/DATABASE_SCHEMA.md)** - Database structure
 - **[docs/BACKEND_STRUCTURE.md](docs/BACKEND_STRUCTURE.md)** - Backend organization
 ### Component Documentation
 - **[docs/CRAWLER_HOW_IT_WORKS.md](docs/CRAWLER_HOW_IT_WORKS.md)** - Crawler internals
 - **[docs/EXTRACTION_STRATEGIES.md](docs/EXTRACTION_STRATEGIES.md)** - Content extraction
 - **[docs/RSS_URL_EXTRACTION.md](docs/RSS_URL_EXTRACTION.md)** - RSS parsing
 ## 🧪 Testing
 All test files are organized in the `tests/` directory:
 ```bash
 # Run crawler tests
 docker-compose exec crawler python tests/crawler/test_crawler.py
 # Run sender tests
 docker-compose exec sender python tests/sender/test_tracking_integration.py
 # Run backend tests
 docker-compose exec backend python tests/backend/test_tracking.py
 ```
 ## 🚀 Production Deployment
 ### Environment Setup
 1. Update `backend/.env` with production values
 2. Set strong MongoDB password
 3. Use HTTPS for tracking URLs
 4. Configure proper SMTP server
 ### Security
 ```bash
 # Use production compose file
 docker-compose -f docker-compose.prod.yml up -d
 # Set MongoDB password
 export MONGO_PASSWORD=your-secure-password
 ```
 ### Monitoring
 - Set up log rotation
 - Configure health checks
 - Set up alerts for failures
 - Monitor database size
 ## 📚 Documentation
 Complete documentation available in the [docs/](docs/) directory:
 - **[Documentation Index](docs/INDEX.md)** - Complete documentation guide
 - **[GPU Setup](docs/GPU_SETUP.md)** - 5-10x faster with GPU acceleration
 - **[Admin API](docs/ADMIN_API.md)** - API endpoints reference
 - **[Security Guide](docs/SECURITY_NOTES.md)** - Security best practices
 - **[System Architecture](docs/SYSTEM_ARCHITECTURE.md)** - Technical overview
 ## 📝 License
 [Your License Here]
 ## 🤝 Contributing
-Contributions welcome! Please read [CONTRIBUTING.md](CONTRIBUTING.md) first.
+We welcome contributions! Please check [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines.
-## 📧 Support
+## 📄 License
-For issues or questions, please open a GitHub issue.
+MIT License - see [LICENSE](LICENSE) for details.
 ---
 **Built with ❤️ for Munich News Daily**
--- a/backend/.env.local
+++ b/backend/.env.local
@@ -0,0 +1,30 @@
 # Munich News Daily - Local Development Backend Configuration
 # MongoDB Configuration
 MONGODB_URI=mongodb://admin:changeme@mongodb:27017/
 # Email Configuration (use test credentials or disable)
 SMTP_SERVER=localhost
 SMTP_PORT=587
 EMAIL_USER=test@localhost
 EMAIL_PASSWORD=test123
 # Newsletter Settings
 NEWSLETTER_MAX_ARTICLES=5
 NEWSLETTER_HOURS_LOOKBACK=24
 WEBSITE_URL=http://localhost:3000
 # Tracking Configuration
 TRACKING_ENABLED=true
 TRACKING_API_URL=http://localhost:5001
 TRACKING_DATA_RETENTION_DAYS=90
 # Ollama Configuration (AI Summarization)
 OLLAMA_ENABLED=true
 OLLAMA_BASE_URL=http://ollama:11434
 OLLAMA_MODEL=phi3:latest
 OLLAMA_TIMEOUT=120
 SUMMARY_MAX_WORDS=150
 # Flask Server Configuration
 FLASK_PORT=5001
--- a/backend/add_categories_to_feeds.py
+++ b/backend/add_categories_to_feeds.py
@@ -0,0 +1,77 @@
 #!/usr/bin/env python
 """
 Script to add categories to existing RSS feeds
 """
 from pymongo import MongoClient
 from config import Config
 client = MongoClient(Config.MONGODB_URI)
 db = client[Config.DB_NAME]
 rss_feeds_collection = db['rss_feeds']
 print("\n" + "="*70)
 print("📡 Adding Categories to RSS Feeds")
 print("="*70)
 # Get all feeds
 all_feeds = list(rss_feeds_collection.find())
 print(f"\nFound {len(all_feeds)} RSS feeds")
 # Category mapping based on feed names/URLs
 category_mapping = {
    'münchen': 'local',
    'munich': 'local',
    'lokales': 'local',
    'sport': 'sports',
    'fussball': 'sports',
    'fc bayern': 'sports',
    'wissenschaft': 'science',
    'science': 'science',
    'tech': 'science',
    'technologie': 'science',
 }
 updated = 0
 for feed in all_feeds:
    name = feed.get('name', '').lower()
    url = feed.get('url', '').lower()
    current_category = feed.get('category')
    # Skip if already has a category
    if current_category:
        print(f"  ✓ {feed['name']}: Already has category '{current_category}'")
        continue
    # Try to determine category from name or URL
    detected_category = 'general'  # Default
    for keyword, category in category_mapping.items():
        if keyword in name or keyword in url:
            detected_category = category
            break
    # Update the feed
    rss_feeds_collection.update_one(
        {'_id': feed['_id']},
        {'$set': {'category': detected_category}}
    )
    print(f"  ✓ {feed['name']}: Set category to '{detected_category}'")
    updated += 1
 print("\n" + "="*70)
 print("📊 Summary")
 print("="*70)
 print(f"✓ Updated: {updated} feeds")
 print(f"✓ Already had categories: {len(all_feeds) - updated} feeds")
 print("="*70 + "\n")
 # Show final category distribution
 print("Category distribution:")
 categories = rss_feeds_collection.aggregate([
    {'$group': {'_id': '$category', 'count': {'$sum': 1}}}
 ])
 for cat in categories:
    print(f"  {cat['_id']}: {cat['count']} feeds")
 print("\n✅ Done! Now run the migration script to update subscriber categories.")
--- a/backend/app.py
+++ b/backend/app.py
@@ -11,6 +11,9 @@ from routes.tracking_routes import tracking_bp
 from routes.analytics_routes import analytics_bp
 from routes.admin_routes import admin_bp
 from routes.transport_routes import transport_bp
 from routes.interests_routes import interests_bp
 from routes.personalization_routes import personalization_bp
 from routes.search_routes import search_bp
 # Initialize Flask app
 app = Flask(__name__)
@@ -29,6 +32,9 @@ app.register_blueprint(tracking_bp)
 app.register_blueprint(analytics_bp)
 app.register_blueprint(admin_bp)
 app.register_blueprint(transport_bp)
 app.register_blueprint(interests_bp)
 app.register_blueprint(personalization_bp)
 app.register_blueprint(search_bp)
 # Health check endpoint
@app.route('/health')
--- a/backend/check_categories.py
+++ b/backend/check_categories.py
@@ -0,0 +1,60 @@
 #!/usr/bin/env python
 """
 Check what categories exist in RSS feeds and articles
 """
 from pymongo import MongoClient
 from config import Config
 client = MongoClient(Config.MONGODB_URI)
 db = client[Config.DB_NAME]
 print("\n" + "="*70)
 print("📊 Category Analysis")
 print("="*70)
 # Check RSS feed categories
 print("\n🔹 RSS Feed Categories:")
 rss_feeds_collection = db['rss_feeds']
 feed_categories = rss_feeds_collection.distinct('category')
 print(f"  Unique categories: {feed_categories}")
 print(f"  Total: {len(feed_categories)} categories")
 # Count feeds per category
 print("\n  Feeds per category:")
 for cat in feed_categories:
    count = rss_feeds_collection.count_documents({'category': cat})
    feeds = list(rss_feeds_collection.find({'category': cat}, {'name': 1, '_id': 0}))
    print(f"    {cat}: {count} feeds")
    for feed in feeds:
        print(f"      - {feed['name']}")
 # Check article categories
 print("\n🔹 Article Categories:")
 articles_collection = db['articles']
 article_categories = articles_collection.distinct('category')
 print(f"  Unique categories: {article_categories}")
 print(f"  Total: {len(article_categories)} categories")
 # Count articles per category
 print("\n  Articles per category:")
 for cat in article_categories:
    count = articles_collection.count_documents({'category': cat})
    with_summary = articles_collection.count_documents({'category': cat, 'summary': {'$exists': True, '$ne': None}})
    print(f"    {cat}: {count} articles ({with_summary} with summaries)")
 # Check subscriber categories
 print("\n🔹 Subscriber Categories:")
 subscribers_collection = db['subscribers']
 total_subs = subscribers_collection.count_documents({'status': 'active'})
 print(f"  Total active subscribers: {total_subs}")
 # Sample a few subscribers to see their categories
 sample_subs = list(subscribers_collection.find({'status': 'active'}, {'email': 1, 'categories': 1, '_id': 0}).limit(5))
 print("\n  Sample subscriber preferences:")
 for sub in sample_subs:
    cats = sub.get('categories', 'None')
    print(f"    {sub['email']}: {cats}")
 print("\n" + "="*70)
 print("✅ Analysis Complete")
 print("="*70 + "\n")
--- a/backend/chroma_client.py
+++ b/backend/chroma_client.py
@@ -0,0 +1,166 @@
 """
 ChromaDB Client for storing and retrieving document embeddings
 """
 import chromadb
 from chromadb.config import Settings
 from chromadb.utils import embedding_functions
 import logging
 import os
 import time
 class ChromaClient:
    """
    Client for interacting with ChromaDB vector database.
    Uses Ollama for generating embeddings if available, otherwise falls back to default.
    """
    def __init__(self, host, port, collection_name='munich_news_articles', ollama_base_url=None):
        """
        Initialize ChromaDB client
        Args:
            host: ChromaDB host (e.g. 'localhost' or 'chromadb')
            port: ChromaDB port (default 8000)
            collection_name: Name of the collection to use
            ollama_base_url: Optional URL for Ollama embedding function
        """
        self.host = host
        self.port = port
        self.collection_name = collection_name
        self.client = None
        self.collection = None
        # Setup embedding function
        # We prefer using a local embedding model compatible with Ollama or SentenceTransformers
        # For simplicity in this stack, we can use the default SentenceTransformer (all-MiniLM-L6-v2)
        # which is downloaded automatically by chromadb utils.
        # Alternatively, we could define a custom function using Ollama's /api/embeddings
        self.embedding_function = embedding_functions.DefaultEmbeddingFunction()
    def connect(self):
        """Establish connection to ChromaDB"""
        try:
            self.client = chromadb.HttpClient(
                host=self.host,
                port=self.port,
                settings=Settings(allow_reset=True, anonymized_telemetry=False)
            )
            # Create or get collection
            self.collection = self.client.get_or_create_collection(
                name=self.collection_name,
                embedding_function=self.embedding_function,
                metadata={"hnsw:space": "cosine"}
            )
            print(f"✓ Connected to ChromaDB at {self.host}:{self.port}")
            return True
        except Exception as e:
            print(f"⚠ Could not connect to ChromaDB: {e}")
            return False
    def add_articles(self, articles):
        """
        Add articles to the vector database
        Args:
            articles: List of dictionaries containing article data.
                     Must have 'link' (used as ID), 'title', 'content', etc.
        """
        if not self.client or not self.collection:
            if not self.connect():
                return False
        if not articles:
            return True
        ids = []
        documents = []
        metadatas = []
        for article in articles:
            # Skip if critical data missing
            if not article.get('link') or not article.get('content'):
                continue
            # Use link as unique ID
            article_id = article.get('link')
            # Prepare text for embedding (Title + Summary + Start of Content)
            # This gives semantic search a good overview
            # Use English title if available, otherwise original
            title = article.get('title_en') if article.get('title_en') else article.get('title', '')
            summary = article.get('summary') or ''
            content_snippet = article.get('content', '')[:1000]
            text_to_embed = f"{title}\n\n{summary}\n\n{content_snippet}"
            # robust metadata (flat dict, no nested objects)
            metadata = {
                "title": title[:100], # Truncate for metadata limits
                "url": article_id,
                "source": article.get('source', 'unknown'),
                "category": article.get('category', 'general'),
                "published_at": str(article.get('published_at', '')),
                "mongo_id": str(article.get('_id', ''))
            }
            ids.append(article_id)
            documents.append(text_to_embed)
            metadatas.append(metadata)
        if not ids:
            return True
        try:
            self.collection.upsert(
                ids=ids,
                documents=documents,
                metadatas=metadatas
            )
            print(f"✓ Indexed {len(ids)} articles in ChromaDB")
            return True
        except Exception as e:
            print(f"✗ Failed to index in ChromaDB: {e}")
            return False
    def search(self, query_text, n_results=5, where=None):
        """
        Search for relevant articles
        Args:
            query_text: The search query
            n_results: Number of results to return
            where: Metadata filter dict (e.g. {"category": "sports"})
        """
        if not self.client or not self.collection:
            if not self.connect():
                return []
        try:
            results = self.collection.query(
                query_texts=[query_text],
                n_results=n_results,
                where=where
            )
            # Format results into a nice list of dicts
            formatted_results = []
            if results and results['ids']:
                for i, id in enumerate(results['ids'][0]):
                    item = {
                        'id': id,
                        'document': results['documents'][0][i] if results['documents'] else None,
                        'metadata': results['metadatas'][0][i] if results['metadatas'] else {},
                        'distance': results['distances'][0][i] if results['distances'] else 0
                    }
                    formatted_results.append(item)
            return formatted_results
        except Exception as e:
            print(f"✗ Search failed: {e}")
            return []
 if __name__ == "__main__":
    # Test client
    client = ChromaClient(host='localhost', port=8000)
    client.connect()
--- a/backend/config.py
+++ b/backend/config.py
@@ -45,6 +45,11 @@ class Config:
    TRACKING_API_URL = os.getenv('TRACKING_API_URL', f'http://localhost:{os.getenv("FLASK_PORT", "5000")}')
    TRACKING_DATA_RETENTION_DAYS = int(os.getenv('TRACKING_DATA_RETENTION_DAYS', '90'))
    # ChromaDB
    CHROMA_HOST = os.getenv('CHROMA_HOST', 'chromadb')
    CHROMA_PORT = int(os.getenv('CHROMA_PORT', '8000'))
    CHROMA_COLLECTION = os.getenv('CHROMA_COLLECTION', 'munich_news_articles')
    @classmethod
    def print_config(cls):
        """Print configuration (without sensitive data)"""
@@ -57,3 +62,5 @@ class Config:
        print(f"  Ollama Enabled: {cls.OLLAMA_ENABLED}")
        print(f"  Tracking Enabled: {cls.TRACKING_ENABLED}")
        print(f"  Tracking API URL: {cls.TRACKING_API_URL}")
        print(f"  ChromaDB Host: {cls.CHROMA_HOST}")
        print(f"  ChromaDB Port: {cls.CHROMA_PORT}")
--- a/backend/migrate_subscriber_categories.py
+++ b/backend/migrate_subscriber_categories.py
@@ -0,0 +1,79 @@
 #!/usr/bin/env python
 """
 Migration script to update subscriber categories
 Ensures all subscribers have all available categories if they're missing some
 """
 from pymongo import MongoClient
 from config import Config
 from database import subscribers_collection, rss_feeds_collection
 def migrate_subscriber_categories():
    """Update all subscribers to have all available categories"""
    print("\n" + "="*70)
    print("📧 Subscriber Categories Migration")
    print("="*70)
    # Get all available categories from RSS feeds
    available_categories = list(rss_feeds_collection.distinct('category'))
    available_categories.sort()
    print(f"\n✓ Available categories: {available_categories}")
    # Get all subscribers
    all_subscribers = list(subscribers_collection.find({}))
    print(f"✓ Found {len(all_subscribers)} total subscribers")
    updated_count = 0
    no_change_count = 0
    for subscriber in all_subscribers:
        email = subscriber['email']
        current_categories = subscriber.get('categories', [])
        # Check if subscriber is missing any categories
        if not current_categories:
            # No categories set - give them all
            print(f"\n  {email}: No categories → Adding all {available_categories}")
            subscribers_collection.update_one(
                {'email': email},
                {'$set': {'categories': available_categories}}
            )
            updated_count += 1
        elif set(current_categories) != set(available_categories):
            # Has some categories but not all
            missing = set(available_categories) - set(current_categories)
            print(f"\n  {email}: {current_categories} → Adding all {available_categories}")
            print(f"    Missing: {list(missing)}")
            subscribers_collection.update_one(
                {'email': email},
                {'$set': {'categories': available_categories}}
            )
            updated_count += 1
        else:
            # Already has all categories
            no_change_count += 1
    print("\n" + "="*70)
    print("📊 Migration Complete")
    print("="*70)
    print(f"✓ Updated: {updated_count} subscribers")
    print(f"✓ No change needed: {no_change_count} subscribers")
    print(f"✓ Total: {len(all_subscribers)} subscribers")
    print("="*70 + "\n")
    return {
        'total': len(all_subscribers),
        'updated': updated_count,
        'no_change': no_change_count
    }
 if __name__ == '__main__':
    try:
        result = migrate_subscriber_categories()
        print(f"✅ Migration successful!")
    except Exception as e:
        print(f"\n❌ Migration failed: {e}")
        import traceback
        traceback.print_exc()
--- a/backend/requirements.txt
+++ b/backend/requirements.txt
@@ -7,3 +7,5 @@ requests==2.31.0
 Jinja2==3.1.2
 redis==5.0.1
 chromadb>=0.4.0
 sentence-transformers>=2.2.2
--- a/backend/routes/interests_routes.py
+++ b/backend/routes/interests_routes.py
@@ -0,0 +1,239 @@
 """
 User Interest Profile API routes for Munich News Daily.
 Provides endpoints to view and manage user interest profiles.
 """
 from flask import Blueprint, request, jsonify
 from services.interest_profiling_service import (
    get_user_interests,
    get_top_interests,
    build_interests_from_history,
    decay_user_interests,
    get_interest_statistics,
    delete_user_interests
 )
 interests_bp = Blueprint('interests', __name__)
@interests_bp.route('/api/interests/<email>', methods=['GET'])
 def get_interests(email):
    """
    Get user interest profile.
    Args:
        email: Email address of the user
    Returns:
        JSON response with user interest profile
    """
    try:
        profile = get_user_interests(email)
        if not profile:
            return jsonify({
                'success': False,
                'error': 'User profile not found'
            }), 404
        # Remove MongoDB _id field
        if '_id' in profile:
            del profile['_id']
        return jsonify({
            'success': True,
            'profile': profile
        }), 200
    except Exception as e:
        return jsonify({
            'success': False,
            'error': str(e)
        }), 500
@interests_bp.route('/api/interests/<email>/top', methods=['GET'])
 def get_top_user_interests(email):
    """
    Get user's top interests sorted by score.
    Query parameters:
        top_n: Number of top interests to return (default: 10)
    Args:
        email: Email address of the user
    Returns:
        JSON response with top categories and keywords
    """
    try:
        top_n = request.args.get('top_n', 10, type=int)
        top_interests = get_top_interests(email, top_n)
        return jsonify({
            'success': True,
            'email': email,
            'top_categories': [
                {'category': cat, 'score': score}
                for cat, score in top_interests['top_categories']
            ],
            'top_keywords': [
                {'keyword': kw, 'score': score}
                for kw, score in top_interests['top_keywords']
            ]
        }), 200
    except Exception as e:
        return jsonify({
            'success': False,
            'error': str(e)
        }), 500
@interests_bp.route('/api/interests/<email>/rebuild', methods=['POST'])
 def rebuild_interests(email):
    """
    Rebuild user interest profile from click history.
    Request body (optional):
        {
            "days_lookback": 30  // Number of days of history to analyze
        }
    Args:
        email: Email address of the user
    Returns:
        JSON response with rebuilt profile
    """
    try:
        data = request.get_json() or {}
        days_lookback = data.get('days_lookback', 30)
        # Validate days_lookback
        if not isinstance(days_lookback, int) or days_lookback < 1:
            return jsonify({
                'success': False,
                'error': 'days_lookback must be a positive integer'
            }), 400
        profile = build_interests_from_history(email, days_lookback)
        # Remove MongoDB _id field
        if '_id' in profile:
            del profile['_id']
        return jsonify({
            'success': True,
            'message': f'Profile rebuilt from {days_lookback} days of history',
            'profile': profile
        }), 200
    except Exception as e:
        return jsonify({
            'success': False,
            'error': str(e)
        }), 500
@interests_bp.route('/api/interests/decay', methods=['POST'])
 def decay_interests():
    """
    Decay interest scores for inactive users.
    Request body (optional):
        {
            "decay_factor": 0.95,      // Multiplier for scores (default: 0.95)
            "days_threshold": 7        // Only decay profiles older than N days
        }
    Returns:
        JSON response with decay statistics
    """
    try:
        data = request.get_json() or {}
        decay_factor = data.get('decay_factor', 0.95)
        days_threshold = data.get('days_threshold', 7)
        # Validate parameters
        if not isinstance(decay_factor, (int, float)) or decay_factor <= 0 or decay_factor > 1:
            return jsonify({
                'success': False,
                'error': 'decay_factor must be between 0 and 1'
            }), 400
        if not isinstance(days_threshold, int) or days_threshold < 1:
            return jsonify({
                'success': False,
                'error': 'days_threshold must be a positive integer'
            }), 400
        result = decay_user_interests(decay_factor, days_threshold)
        return jsonify({
            'success': True,
            'message': f'Decayed interests for profiles older than {days_threshold} days',
            'statistics': result
        }), 200
    except Exception as e:
        return jsonify({
            'success': False,
            'error': str(e)
        }), 500
@interests_bp.route('/api/interests/statistics', methods=['GET'])
 def get_statistics():
    """
    Get statistics about user interests across all users.
    Returns:
        JSON response with interest statistics
    """
    try:
        stats = get_interest_statistics()
        return jsonify({
            'success': True,
            'statistics': stats
        }), 200
    except Exception as e:
        return jsonify({
            'success': False,
            'error': str(e)
        }), 500
@interests_bp.route('/api/interests/<email>', methods=['DELETE'])
 def delete_interests(email):
    """
    Delete user interest profile (GDPR compliance).
    Args:
        email: Email address of the user
    Returns:
        JSON response with confirmation
    """
    try:
        deleted = delete_user_interests(email)
        if not deleted:
            return jsonify({
                'success': False,
                'error': 'User profile not found'
            }), 404
        return jsonify({
            'success': True,
            'message': f'Interest profile deleted for {email}'
        }), 200
    except Exception as e:
        return jsonify({
            'success': False,
            'error': str(e)
        }), 500
--- a/backend/routes/news_routes.py
+++ b/backend/routes/news_routes.py
@@ -24,8 +24,11 @@ def get_news():
        db_articles = []
        for doc in cursor:
            # Use English title if available, otherwise fallback to original
            title = doc.get('title_en') if doc.get('title_en') else doc.get('title', '')
            article = {
-                'title': doc.get('title', ''),
+                'title': title,
                'author': doc.get('author'),
                'link': doc.get('link', ''),
                'source': doc.get('source', ''),
@@ -114,8 +117,10 @@ def get_clustered_news_internal():
            # Use cluster_articles from aggregation (already fetched)
            cluster_articles = doc.get('cluster_articles', [])
            title = doc.get('title_en') if doc.get('title_en') else doc.get('title', '')
            article = {
-                'title': doc.get('title', ''),
+                'title': title,
                'link': doc.get('link', ''),
                'source': doc.get('source', ''),
                'published': doc.get('published_at', ''),
@@ -173,7 +178,7 @@ def get_article_by_url(article_url):
            return jsonify({'error': 'Article not found'}), 404
        return jsonify({
-            'title': article.get('title', ''),
+            'title': article.get('title_en') if article.get('title_en') else article.get('title', ''),
            'author': article.get('author'),
            'link': article.get('link', ''),
            'content': article.get('content', ''),
--- a/backend/routes/personalization_routes.py
+++ b/backend/routes/personalization_routes.py
@@ -0,0 +1,135 @@
 """
 Personalization API routes for Munich News Daily.
 Provides endpoints to test and preview personalized content.
 """
 from flask import Blueprint, request, jsonify
 from datetime import datetime, timedelta
 from database import articles_collection
 from services.personalization_service import (
    rank_articles_for_user,
    select_personalized_articles,
    get_personalization_explanation,
    get_personalization_stats
 )
 personalization_bp = Blueprint('personalization', __name__)
@personalization_bp.route('/api/personalize/preview/<email>', methods=['GET'])
 def preview_personalized_newsletter(email):
    """
    Preview personalized newsletter for a user.
    Query parameters:
        max_articles: Maximum articles to return (default: 10)
        hours_lookback: Hours of articles to consider (default: 24)
    Returns:
        JSON with personalized article selection and statistics
    """
    try:
        max_articles = request.args.get('max_articles', 10, type=int)
        hours_lookback = request.args.get('hours_lookback', 24, type=int)
        # Get recent articles
        cutoff_date = datetime.utcnow() - timedelta(hours=hours_lookback)
        articles = list(articles_collection.find({
            'created_at': {'$gte': cutoff_date},
            'summary': {'$exists': True, '$ne': None}
        }).sort('created_at', -1))
        # Select personalized articles
        personalized = select_personalized_articles(
            articles,
            email,
            max_articles=max_articles
        )
        # Get statistics
        stats = get_personalization_stats(personalized, email)
        # Format response
        articles_response = []
        for article in personalized:
            articles_response.append({
                'title': article.get('title', ''),
                'title_en': article.get('title_en'),
                'summary': article.get('summary', ''),
                'link': article.get('link', ''),
                'category': article.get('category', 'general'),
                'keywords': article.get('keywords', []),
                'personalization_score': article.get('personalization_score', 0.0),
                'published_at': article.get('published_at', '')
            })
        return jsonify({
            'success': True,
            'email': email,
            'articles': articles_response,
            'statistics': stats
        }), 200
    except Exception as e:
        return jsonify({
            'success': False,
            'error': str(e)
        }), 500
@personalization_bp.route('/api/personalize/explain', methods=['POST'])
 def explain_recommendation():
    """
    Explain why an article was recommended to a user.
    Request body:
        {
            "email": "user@example.com",
            "article_id": "article-id-here"
        }
    Returns:
        JSON with explanation of recommendation
    """
    try:
        data = request.get_json()
        if not data or 'email' not in data or 'article_id' not in data:
            return jsonify({
                'success': False,
                'error': 'email and article_id required'
            }), 400
        email = data['email']
        article_id = data['article_id']
        # Get article
        from bson import ObjectId
        article = articles_collection.find_one({'_id': ObjectId(article_id)})
        if not article:
            return jsonify({
                'success': False,
                'error': 'Article not found'
            }), 404
        # Get user interests
        from services.interest_profiling_service import get_user_interests
        user_interests = get_user_interests(email)
        # Generate explanation
        explanation = get_personalization_explanation(article, user_interests)
        return jsonify({
            'success': True,
            'email': email,
            'article_title': article.get('title', ''),
            'explanation': explanation
        }), 200
    except Exception as e:
        return jsonify({
            'success': False,
            'error': str(e)
        }), 500
--- a/backend/routes/search_routes.py
+++ b/backend/routes/search_routes.py
@@ -0,0 +1,88 @@
 from flask import Blueprint, jsonify, request
 from config import Config
 from chroma_client import ChromaClient
 import logging
 search_bp = Blueprint('search', __name__)
 # Initialize ChromaDB client
 # Note: We use the hostname 'chromadb' as defined in docker-compose for the backend
 chroma_client = ChromaClient(
    host=Config.CHROMA_HOST,
    port=Config.CHROMA_PORT,
    collection_name=Config.CHROMA_COLLECTION
 )
@search_bp.route('/api/search', methods=['GET'])
 def search_news():
    """
    Semantic search for news articles using ChromaDB.
    Query parameters:
    - q: Search query (required)
    - limit: Number of results (default: 10)
    - category: Filter by category (optional)
    """
    try:
        query = request.args.get('q')
        if not query:
            return jsonify({'error': 'Missing search query'}), 400
        limit = int(request.args.get('limit', 10))
        category = request.args.get('category')
        # Build filter if category provided
        where_filter = None
        if category:
            where_filter = {"category": category}
        # Perform search
        results = chroma_client.search(
            query_text=query,
            n_results=limit,
            where=where_filter
        )
        # Format for frontend
        formatted_response = []
        for item in results:
            metadata = item.get('metadata', {})
            # Use translated title if availble (stored in metadata as title_en or title)
            # Note: Chroma metadata structure is flat. If we store title_en, we should use it.
            # But currently we store: title, url, source, category, published_at. 
            # We need to make sure title_en is stored in Chroma OR fetch it from DB.
            # Faster approach: just rely on what is in Chroma.
            # BETTER: In crawl, we store title as title_en in metadata if available?
            # Let's check how we store it in crawler_service.py/chroma_client.py
            # Correction: Looking at crawler_service.py line 456, we pass article_doc to add_articles.
            # In chroma_client.py line 97, we only extract title, url, source, category, published_at.
            # We are NOT storing title_en in Chroma metadata currently.
            # FOR NOW: We will stick to the title stored in Chroma, but we should update Chroma storing logic.
            # However, since the user IS complaining about English, let's assume valid English titles 
            # are what we want to display.
            # Wait, if we change the metadata in ChromaClient to use title_en as the main title, 
            # then search results will automatically show English.
            title = metadata.get('title', 'Unknown Title')
            formatted_response.append({
                'title': title,
                'link': metadata.get('url', ''),
                'source': metadata.get('source', 'Unknown'),
                'category': metadata.get('category', 'general'),
                'published_at': metadata.get('published_at', ''),
                'relevance_score': 1.0 - item.get('distance', 1.0), # Convert distance to score (approx)
                'snippet': item.get('document', '')[:200] + '...' # Preview
            })
        return jsonify({
            'query': query,
            'count': len(formatted_response),
            'results': formatted_response
        }), 200
    except Exception as e:
        logging.error(f"Search error: {str(e)}")
        return jsonify({'error': str(e)}), 500
--- a/backend/routes/tracking_routes.py
+++ b/backend/routes/tracking_routes.py
@@ -79,8 +79,8 @@ def track_click(tracking_id):
    """
    Track link clicks and redirect to original article URL.
-    Logs the click event and redirects the user to the original article URL.
+    Logs the click event, updates user interest profile, and redirects the user
-    Handles invalid tracking_id by redirecting to homepage.
+    to the original article URL. Handles invalid tracking_id by redirecting to homepage.
    Ensures redirect completes within 200ms.
    Args:
@@ -115,6 +115,19 @@ def track_click(tracking_id):
                    }
                }
            )
            # Update user interest profile (Phase 3)
            subscriber_email = tracking_record.get('subscriber_email')
            keywords = tracking_record.get('keywords', [])
            category = tracking_record.get('category', 'general')
            if subscriber_email and subscriber_email != 'anonymized':
                try:
                    from services.interest_profiling_service import update_user_interests
                    update_user_interests(subscriber_email, keywords, category)
                except Exception as e:
                    # Don't fail the redirect if interest update fails
                    print(f"Error updating user interests: {str(e)}")
    except Exception as e:
        # Log error but still redirect
        print(f"Error tracking click for {tracking_id}: {str(e)}")
--- a/backend/services/interest_profiling_service.py
+++ b/backend/services/interest_profiling_service.py
@@ -0,0 +1,323 @@
 """
 User Interest Profiling Service for Munich News Daily.
 Builds and maintains user interest profiles based on article click behavior.
 """
 from datetime import datetime, timedelta
 from typing import Dict, List, Optional
 from database import link_clicks_collection
 from pymongo import MongoClient
 from config import Config
 # Connect to MongoDB
 client = MongoClient(Config.MONGODB_URI)
 db = client[Config.DB_NAME]
 user_interests_collection = db['user_interests']
 def update_user_interests(subscriber_email: str, keywords: List[str], category: str) -> Dict:
    """
    Update user interest profile based on a clicked article.
    Increments interest scores for the article's keywords and category.
    Creates a new profile if the user doesn't have one yet.
    Args:
        subscriber_email: Email address of the user
        keywords: List of keywords from the clicked article
        category: Category of the clicked article
    Returns:
        dict: Updated user interest profile
    """
    current_time = datetime.utcnow()
    # Get existing profile or create new one
    profile = user_interests_collection.find_one({'email': subscriber_email})
    if not profile:
        # Create new profile
        profile = {
            'email': subscriber_email,
            'categories': {},
            'keywords': {},
            'total_clicks': 0,
            'last_updated': current_time,
            'created_at': current_time
        }
    # Update category interest (increment by 0.1, max 1.0)
    current_category_score = profile['categories'].get(category, 0.0)
    profile['categories'][category] = min(current_category_score + 0.1, 1.0)
    # Update keyword interests (increment by 0.1, max 1.0)
    for keyword in keywords:
        if keyword:  # Skip empty keywords
            current_keyword_score = profile['keywords'].get(keyword, 0.0)
            profile['keywords'][keyword] = min(current_keyword_score + 0.1, 1.0)
    # Update metadata
    profile['total_clicks'] = profile.get('total_clicks', 0) + 1
    profile['last_updated'] = current_time
    # Upsert profile
    user_interests_collection.update_one(
        {'email': subscriber_email},
        {'$set': profile},
        upsert=True
    )
    return profile
 def get_user_interests(subscriber_email: str) -> Optional[Dict]:
    """
    Get user interest profile.
    Args:
        subscriber_email: Email address of the user
    Returns:
        dict: User interest profile or None if not found
    """
    return user_interests_collection.find_one({'email': subscriber_email})
 def decay_user_interests(decay_factor: float = 0.95, days_threshold: int = 7) -> Dict[str, int]:
    """
    Decay interest scores for users who haven't clicked recently.
    Reduces interest scores over time to reflect changing interests.
    Only decays profiles that haven't been updated in the last N days.
    Args:
        decay_factor: Multiplier for interest scores (default: 0.95 = 5% decay)
        days_threshold: Only decay profiles older than this many days (default: 7)
    Returns:
        dict: Statistics about the decay operation
            - profiles_decayed: Number of profiles that were decayed
            - profiles_checked: Total number of profiles checked
    """
    cutoff_date = datetime.utcnow() - timedelta(days=days_threshold)
    # Find profiles that haven't been updated recently
    old_profiles = user_interests_collection.find({
        'last_updated': {'$lt': cutoff_date}
    })
    profiles_decayed = 0
    profiles_checked = 0
    for profile in old_profiles:
        profiles_checked += 1
        # Decay category scores
        decayed_categories = {}
        for category, score in profile.get('categories', {}).items():
            new_score = score * decay_factor
            # Remove categories with very low scores (< 0.05)
            if new_score >= 0.05:
                decayed_categories[category] = round(new_score, 3)
        # Decay keyword scores
        decayed_keywords = {}
        for keyword, score in profile.get('keywords', {}).items():
            new_score = score * decay_factor
            # Remove keywords with very low scores (< 0.05)
            if new_score >= 0.05:
                decayed_keywords[keyword] = round(new_score, 3)
        # Update profile with decayed scores
        user_interests_collection.update_one(
            {'email': profile['email']},
            {
                '$set': {
                    'categories': decayed_categories,
                    'keywords': decayed_keywords,
                    'last_decayed': datetime.utcnow()
                }
            }
        )
        profiles_decayed += 1
    return {
        'profiles_decayed': profiles_decayed,
        'profiles_checked': profiles_checked
    }
 def get_top_interests(subscriber_email: str, top_n: int = 10) -> Dict[str, List[tuple]]:
    """
    Get user's top interests sorted by score.
    Args:
        subscriber_email: Email address of the user
        top_n: Number of top interests to return (default: 10)
    Returns:
        dict: Top interests containing:
            - top_categories: List of (category, score) tuples
            - top_keywords: List of (keyword, score) tuples
    """
    profile = get_user_interests(subscriber_email)
    if not profile:
        return {
            'top_categories': [],
            'top_keywords': []
        }
    # Sort categories by score
    categories = profile.get('categories', {})
    top_categories = sorted(categories.items(), key=lambda x: x[1], reverse=True)[:top_n]
    # Sort keywords by score
    keywords = profile.get('keywords', {})
    top_keywords = sorted(keywords.items(), key=lambda x: x[1], reverse=True)[:top_n]
    return {
        'top_categories': top_categories,
        'top_keywords': top_keywords
    }
 def build_interests_from_history(subscriber_email: str, days_lookback: int = 30) -> Dict:
    """
    Build or rebuild user interest profile from click history.
    Useful for:
    - Initializing profiles for existing users
    - Rebuilding profiles after algorithm changes
    - Backfilling data
    Args:
        subscriber_email: Email address of the user
        days_lookback: Number of days of history to analyze (default: 30)
    Returns:
        dict: Newly built interest profile
    """
    cutoff_date = datetime.utcnow() - timedelta(days=days_lookback)
    # Get all clicks from this user in the lookback period
    clicks = link_clicks_collection.find({
        'subscriber_email': subscriber_email,
        'clicked': True,
        'clicked_at': {'$gte': cutoff_date}
    })
    # Initialize profile
    profile = {
        'email': subscriber_email,
        'categories': {},
        'keywords': {},
        'total_clicks': 0,
        'last_updated': datetime.utcnow(),
        'created_at': datetime.utcnow()
    }
    # Process each click
    for click in clicks:
        category = click.get('category', 'general')
        keywords = click.get('keywords', [])
        # Update category score
        profile['categories'][category] = profile['categories'].get(category, 0.0) + 0.1
        # Update keyword scores
        for keyword in keywords:
            if keyword:
                profile['keywords'][keyword] = profile['keywords'].get(keyword, 0.0) + 0.1
        profile['total_clicks'] += 1
    # Cap scores at 1.0
    for category in profile['categories']:
        profile['categories'][category] = min(profile['categories'][category], 1.0)
    for keyword in profile['keywords']:
        profile['keywords'][keyword] = min(profile['keywords'][keyword], 1.0)
    # Save profile
    if profile['total_clicks'] > 0:
        user_interests_collection.update_one(
            {'email': subscriber_email},
            {'$set': profile},
            upsert=True
        )
    return profile
 def get_interest_statistics() -> Dict:
    """
    Get statistics about user interests across all users.
    Returns:
        dict: Statistics containing:
            - total_users: Total number of users with profiles
            - avg_clicks_per_user: Average number of clicks per user
            - most_popular_categories: Top categories across all users
            - most_popular_keywords: Top keywords across all users
    """
    total_users = user_interests_collection.count_documents({})
    if total_users == 0:
        return {
            'total_users': 0,
            'avg_clicks_per_user': 0,
            'most_popular_categories': [],
            'most_popular_keywords': []
        }
    # Calculate average clicks
    pipeline = [
        {
            '$group': {
                '_id': None,
                'total_clicks': {'$sum': '$total_clicks'}
            }
        }
    ]
    result = list(user_interests_collection.aggregate(pipeline))
    total_clicks = result[0]['total_clicks'] if result else 0
    avg_clicks = total_clicks / total_users if total_users > 0 else 0
    # Get most popular categories
    category_counts = {}
    keyword_counts = {}
    for profile in user_interests_collection.find({}):
        for category, score in profile.get('categories', {}).items():
            category_counts[category] = category_counts.get(category, 0) + score
        for keyword, score in profile.get('keywords', {}).items():
            keyword_counts[keyword] = keyword_counts.get(keyword, 0) + score
    # Sort and get top 10
    top_categories = sorted(category_counts.items(), key=lambda x: x[1], reverse=True)[:10]
    top_keywords = sorted(keyword_counts.items(), key=lambda x: x[1], reverse=True)[:10]
    return {
        'total_users': total_users,
        'avg_clicks_per_user': round(avg_clicks, 2),
        'most_popular_categories': top_categories,
        'most_popular_keywords': top_keywords
    }
 def delete_user_interests(subscriber_email: str) -> bool:
    """
    Delete user interest profile (for GDPR compliance).
    Args:
        subscriber_email: Email address of the user
    Returns:
        bool: True if profile was deleted, False if not found
    """
    result = user_interests_collection.delete_one({'email': subscriber_email})
    return result.deleted_count > 0
--- a/backend/services/personalization_service.py
+++ b/backend/services/personalization_service.py
@@ -0,0 +1,295 @@
 """
 Newsletter Personalization Service for Munich News Daily.
 Ranks and selects articles based on user interest profiles.
 """
 from typing import Dict, List, Optional
 from datetime import datetime, timedelta
 from services.interest_profiling_service import get_user_interests
 def calculate_article_score(
    article: Dict,
    user_interests: Optional[Dict],
    category_weight: float = 0.4,
    keyword_weight: float = 0.6
 ) -> float:
    """
    Calculate personalization score for an article based on user interests.
    Score is calculated as:
    - Category match: 0-1.0 based on user's interest in the category
    - Keyword match: Average of user's interest in article keywords
    - Final score: (category_score * 0.4) + (keyword_score * 0.6)
    Args:
        article: Article dictionary with 'category' and 'keywords' fields
        user_interests: User interest profile (None for non-personalized)
        category_weight: Weight for category matching (default: 0.4)
        keyword_weight: Weight for keyword matching (default: 0.6)
    Returns:
        float: Personalization score between 0.0 and 1.0
    """
    # If no user interests, return neutral score
    if not user_interests:
        return 0.5
    # Get article metadata
    article_category = article.get('category', 'general')
    article_keywords = article.get('keywords', [])
    # Calculate category score
    user_categories = user_interests.get('categories', {})
    category_score = user_categories.get(article_category, 0.0)
    # Calculate keyword score (average of all matching keywords)
    user_keywords = user_interests.get('keywords', {})
    keyword_scores = []
    for keyword in article_keywords:
        if keyword in user_keywords:
            keyword_scores.append(user_keywords[keyword])
    # Average keyword score (0.0 if no matches)
    keyword_score = sum(keyword_scores) / len(keyword_scores) if keyword_scores else 0.0
    # Weighted final score
    final_score = (category_score * category_weight) + (keyword_score * keyword_weight)
    return round(final_score, 3)
 def rank_articles_for_user(
    articles: List[Dict],
    subscriber_email: str,
    personalization_ratio: float = 0.7
 ) -> List[Dict]:
    """
    Rank articles for a specific user based on their interests.
    Mixes personalized content with trending content to avoid filter bubbles.
    Args:
        articles: List of article dictionaries
        subscriber_email: Email address of the user
        personalization_ratio: Ratio of personalized vs trending (default: 0.7 = 70% personalized)
    Returns:
        list: Articles sorted by personalization score with score added
    """
    # Get user interests
    user_interests = get_user_interests(subscriber_email)
    # Calculate score for each article
    scored_articles = []
    for article in articles:
        score = calculate_article_score(article, user_interests)
        # Add score to article (don't modify original)
        article_with_score = article.copy()
        article_with_score['personalization_score'] = score
        scored_articles.append(article_with_score)
    # Sort by score (highest first)
    scored_articles.sort(key=lambda x: x['personalization_score'], reverse=True)
    return scored_articles
 def select_personalized_articles(
    articles: List[Dict],
    subscriber_email: str,
    max_articles: int = 10,
    personalization_ratio: float = 0.7,
    min_score_threshold: float = 0.1
 ) -> List[Dict]:
    """
    Select and rank articles for a personalized newsletter.
    Strategy:
    - Top N * personalization_ratio articles: Highest scoring (personalized)
    - Remaining articles: Most recent (trending/diverse content)
    - Ensures mix of personalized + diverse content
    Args:
        articles: List of available articles
        subscriber_email: Email address of the user
        max_articles: Maximum number of articles to include (default: 10)
        personalization_ratio: Ratio of personalized content (default: 0.7)
        min_score_threshold: Minimum score to consider personalized (default: 0.1)
    Returns:
        list: Selected articles with personalization scores
    """
    if not articles:
        return []
    # Rank all articles
    ranked_articles = rank_articles_for_user(articles, subscriber_email, personalization_ratio)
    # Calculate split
    num_personalized = int(max_articles * personalization_ratio)
    num_trending = max_articles - num_personalized
    # Get personalized articles (high scoring)
    personalized = [
        a for a in ranked_articles 
        if a['personalization_score'] >= min_score_threshold
    ][:num_personalized]
    # Get trending articles (most recent, not already selected)
    personalized_ids = {a.get('_id') for a in personalized}
    trending = [
        a for a in ranked_articles 
        if a.get('_id') not in personalized_ids
    ][:num_trending]
    # Combine: personalized first, then trending
    selected = personalized + trending
    # Ensure we don't exceed max_articles
    return selected[:max_articles]
 def get_personalization_explanation(
    article: Dict,
    user_interests: Optional[Dict]
 ) -> Dict[str, any]:
    """
    Generate explanation for why an article was recommended.
    Useful for transparency and debugging.
    Args:
        article: Article dictionary
        user_interests: User interest profile
    Returns:
        dict: Explanation containing:
            - score: Overall personalization score
            - category_match: Category score
            - keyword_matches: List of matching keywords with scores
            - reason: Human-readable explanation
    """
    if not user_interests:
        return {
            'score': 0.5,
            'category_match': 0.0,
            'keyword_matches': [],
            'reason': 'No personalization data available'
        }
    article_category = article.get('category', 'general')
    article_keywords = article.get('keywords', [])
    user_categories = user_interests.get('categories', {})
    user_keywords = user_interests.get('keywords', {})
    # Category match
    category_score = user_categories.get(article_category, 0.0)
    # Keyword matches
    keyword_matches = []
    for keyword in article_keywords:
        if keyword in user_keywords:
            keyword_matches.append({
                'keyword': keyword,
                'score': user_keywords[keyword]
            })
    # Calculate overall score
    overall_score = calculate_article_score(article, user_interests)
    # Generate reason
    if overall_score >= 0.5:
        reason = f"High match with your interests in {article_category}"
        if keyword_matches:
            top_keywords = [m['keyword'] for m in keyword_matches[:2]]
            reason += f" and topics like {', '.join(top_keywords)}"
    elif overall_score >= 0.3:
        reason = f"Moderate match with your interests"
    else:
        reason = "Trending article for diverse content"
    return {
        'score': overall_score,
        'category_match': category_score,
        'keyword_matches': keyword_matches,
        'reason': reason
    }
 def get_personalization_stats(
    selected_articles: List[Dict],
    subscriber_email: str
 ) -> Dict[str, any]:
    """
    Get statistics about personalization for a newsletter.
    Args:
        selected_articles: Articles selected for the newsletter
        subscriber_email: Email address of the user
    Returns:
        dict: Statistics containing:
            - total_articles: Number of articles
            - avg_score: Average personalization score
            - highly_personalized: Number of articles with score >= 0.5
            - moderately_personalized: Number with score 0.3-0.5
            - trending: Number with score < 0.3
    """
    if not selected_articles:
        return {
            'total_articles': 0,
            'avg_score': 0.0,
            'highly_personalized': 0,
            'moderately_personalized': 0,
            'trending': 0
        }
    scores = [a.get('personalization_score', 0.0) for a in selected_articles]
    avg_score = sum(scores) / len(scores)
    highly_personalized = sum(1 for s in scores if s >= 0.5)
    moderately_personalized = sum(1 for s in scores if 0.3 <= s < 0.5)
    trending = sum(1 for s in scores if s < 0.3)
    return {
        'total_articles': len(selected_articles),
        'avg_score': round(avg_score, 3),
        'highly_personalized': highly_personalized,
        'moderately_personalized': moderately_personalized,
        'trending': trending
    }
 def batch_personalize_newsletters(
    articles: List[Dict],
    subscribers: List[str],
    max_articles_per_user: int = 10
 ) -> Dict[str, List[Dict]]:
    """
    Generate personalized article selections for multiple subscribers.
    Useful for batch newsletter generation.
    Args:
        articles: List of available articles
        subscribers: List of subscriber email addresses
        max_articles_per_user: Max articles per newsletter (default: 10)
    Returns:
        dict: Mapping of email -> personalized article list
    """
    personalized_newsletters = {}
    for subscriber_email in subscribers:
        personalized_articles = select_personalized_articles(
            articles,
            subscriber_email,
            max_articles=max_articles_per_user
        )
        personalized_newsletters[subscriber_email] = personalized_articles
    return personalized_newsletters
--- a/backend/services/tracking_service.py
+++ b/backend/services/tracking_service.py
@@ -80,6 +80,9 @@ def create_newsletter_tracking(
    link_tracking_map = {}
    if article_links:
        # Import here to avoid circular dependency
        from database import articles_collection
        for article in article_links:
            article_url = article.get('url')
            article_title = article.get('title', '')
@@ -87,13 +90,22 @@ def create_newsletter_tracking(
            if article_url:
                link_tracking_id = generate_tracking_id()
-                # Create link click tracking record
+                # Look up article metadata from database for personalization
                article_doc = articles_collection.find_one({'link': article_url})
                article_id = str(article_doc['_id']) if article_doc else None
                category = article_doc.get('category', 'general') if article_doc else 'general'
                keywords = article_doc.get('keywords', []) if article_doc else []
                # Create link click tracking record with metadata
                link_click_doc = {
                    'tracking_id': link_tracking_id,
                    'newsletter_id': newsletter_id,
                    'subscriber_email': subscriber_email,
                    'article_url': article_url,
                    'article_title': article_title,
                    'article_id': article_id,           # NEW: Article database ID
                    'category': category,                # NEW: Article category
                    'keywords': keywords,                # NEW: Article keywords for personalization
                    'clicked': False,
                    'clicked_at': None,
                    'user_agent': None,
--- a/backend/test_personalization_system.py
+++ b/backend/test_personalization_system.py
@@ -0,0 +1,221 @@
 #!/usr/bin/env python3
 """
 Comprehensive test suite for the personalization system.
 Tests all 4 phases: keyword extraction, click tracking, interest profiling, and personalization.
 """
 import sys
 from pymongo import MongoClient
 from datetime import datetime
 # Import services
 from services.tracking_service import create_newsletter_tracking
 from services.interest_profiling_service import (
    update_user_interests,
    get_user_interests,
    get_top_interests,
    build_interests_from_history
 )
 from services.personalization_service import (
    calculate_article_score,
    rank_articles_for_user,
    select_personalized_articles,
    get_personalization_stats
 )
 from config import Config
 # Connect to MongoDB
 client = MongoClient(Config.MONGODB_URI)
 db = client[Config.DB_NAME]
 articles_collection = db['articles']
 link_clicks_collection = db['link_clicks']
 user_interests_collection = db['user_interests']
 def test_phase1_keywords():
    """Phase 1: Verify articles have keywords extracted"""
    print("\n" + "="*60)
    print("Phase 1: Keyword Extraction")
    print("="*60)
    articles_with_keywords = articles_collection.count_documents({
        'keywords': {'$exists': True, '$ne': []}
    })
    if articles_with_keywords == 0:
        print("❌ No articles with keywords found")
        print("   Run a crawl first to extract keywords")
        return False
    sample = articles_collection.find_one({'keywords': {'$exists': True, '$ne': []}})
    print(f"✓ Found {articles_with_keywords} articles with keywords")
    print(f"  Sample: {sample.get('title', 'N/A')[:50]}...")
    print(f"  Keywords: {sample.get('keywords', [])[:3]}")
    return True
 def test_phase2_tracking():
    """Phase 2: Verify tracking includes keywords and metadata"""
    print("\n" + "="*60)
    print("Phase 2: Click Tracking Enhancement")
    print("="*60)
    test_email = 'test-phase2@example.com'
    # Clean up
    link_clicks_collection.delete_many({'subscriber_email': test_email})
    # Get article with keywords
    article = articles_collection.find_one({'keywords': {'$exists': True, '$ne': []}})
    if not article:
        print("❌ No articles found")
        return False
    # Create tracking
    tracking_data = create_newsletter_tracking(
        newsletter_id='test-phase2',
        subscriber_email=test_email,
        article_links=[{
            'url': article['link'],
            'title': article.get('title', '')
        }]
    )
    # Verify tracking record
    tracking_id = list(tracking_data['link_tracking_map'].values())[0]
    tracking_record = link_clicks_collection.find_one({'tracking_id': tracking_id})
    has_metadata = (
        tracking_record.get('article_id') is not None and
        tracking_record.get('category') is not None and
        len(tracking_record.get('keywords', [])) > 0
    )
    # Clean up
    link_clicks_collection.delete_many({'subscriber_email': test_email})
    db['newsletter_sends'].delete_many({'subscriber_email': test_email})
    if has_metadata:
        print(f"✓ Tracking records include metadata")
        print(f"  Article ID: {tracking_record.get('article_id')}")
        print(f"  Category: {tracking_record.get('category')}")
        print(f"  Keywords: {len(tracking_record.get('keywords', []))} keywords")
        return True
    else:
        print("❌ Tracking records missing metadata")
        return False
 def test_phase3_profiling():
    """Phase 3: Verify interest profiles are built from clicks"""
    print("\n" + "="*60)
    print("Phase 3: User Interest Profiling")
    print("="*60)
    test_email = 'test-phase3@example.com'
    # Clean up
    user_interests_collection.delete_many({'email': test_email})
    # Create profile
    update_user_interests(test_email, ['Bayern Munich', 'Football'], 'sports')
    update_user_interests(test_email, ['Transportation', 'Munich'], 'local')
    # Verify profile
    profile = get_user_interests(test_email)
    # Clean up
    user_interests_collection.delete_many({'email': test_email})
    if profile and profile['total_clicks'] == 2:
        print(f"✓ Interest profile created")
        print(f"  Total clicks: {profile['total_clicks']}")
        print(f"  Categories: {len(profile.get('categories', {}))}")
        print(f"  Keywords: {len(profile.get('keywords', {}))}")
        return True
    else:
        print("❌ Interest profile not created correctly")
        return False
 def test_phase4_personalization():
    """Phase 4: Verify articles are ranked by user interests"""
    print("\n" + "="*60)
    print("Phase 4: Personalized Newsletter Generation")
    print("="*60)
    test_email = 'test-phase4@example.com'
    # Clean up
    user_interests_collection.delete_many({'email': test_email})
    # Get articles
    articles = list(articles_collection.find(
        {'keywords': {'$exists': True, '$ne': []}},
        limit=5
    ))
    if len(articles) < 3:
        print("❌ Not enough articles found")
        return False
    # Create profile
    update_user_interests(test_email, ['Bayern Munich', 'Football'], 'sports')
    # Rank articles
    ranked = rank_articles_for_user(articles, test_email)
    # Select personalized
    selected = select_personalized_articles(articles, test_email, max_articles=3)
    # Clean up
    user_interests_collection.delete_many({'email': test_email})
    has_scores = all('personalization_score' in a for a in selected)
    if has_scores and len(selected) > 0:
        print(f"✓ Articles ranked and selected")
        print(f"  Total ranked: {len(ranked)}")
        print(f"  Selected: {len(selected)}")
        print(f"  Top score: {selected[0].get('personalization_score', 0):.3f}")
        return True
    else:
        print("❌ Personalization failed")
        return False
 def main():
    """Run all personalization tests"""
    print("\n" + "="*60)
    print("PERSONALIZATION SYSTEM TEST SUITE")
    print("="*60)
    results = {
        'Phase 1: Keyword Extraction': test_phase1_keywords(),
        'Phase 2: Click Tracking': test_phase2_tracking(),
        'Phase 3: Interest Profiling': test_phase3_profiling(),
        'Phase 4: Personalization': test_phase4_personalization()
    }
    print("\n" + "="*60)
    print("TEST RESULTS")
    print("="*60)
    for phase, passed in results.items():
        status = "✅ PASS" if passed else "❌ FAIL"
        print(f"{status} - {phase}")
    all_passed = all(results.values())
    if all_passed:
        print("\n🎉 All personalization tests PASSED!")
        return 0
    else:
        print("\n❌ Some tests FAILED")
        return 1
 if __name__ == '__main__':
    sys.exit(main())
--- a/debug_categories.py
+++ b/debug_categories.py
@@ -0,0 +1,35 @@
 from pymongo import MongoClient
 from datetime import datetime, timedelta
 import os
 # Connect to MongoDB
 mongo_uri = os.getenv('MONGODB_URI', 'mongodb://mongodb:27017/')
 client = MongoClient(mongo_uri)
 db = client['munich_news']
 articles = db['articles']
 subscribers = db['subscribers']
 print("--- Distinct Categories in Articles Collection ---")
 categories = articles.distinct('category')
 print(categories)
 print("\n--- Recent Article Counts by Category (Last 24h) ---")
 yesterday = datetime.utcnow() - timedelta(hours=24)
 recent_articles = articles.find({'created_at': {'$gte': yesterday}})
 category_counts = {}
 for art in recent_articles:
    cat = art.get('category', 'unknown')
    category_counts[cat] = category_counts.get(cat, 0) + 1
 for cat, count in category_counts.items():
    print(f"{cat}: {count}")
 print("\n--- Subscriber Preferences ---")
 for sub in subscribers.find():
    print(f"Email: {sub.get('email')}, Categories: {sub.get('categories')}")
 print("\n--- RSS Feeds ---")
 rss_feeds = db['rss_feeds']
 for feed in rss_feeds.find():
    print(f"Name: {feed.get('name')}, URL: {feed.get('url')}, Category: {feed.get('category')}, Active: {feed.get('active')}")
--- a/docker-compose.local.yml
+++ b/docker-compose.local.yml
@@ -0,0 +1,225 @@
 services:
  # Ollama AI Service (Exposed for local testing)
  ollama:
    image: ollama/ollama:latest
    container_name: munich-news-local-ollama
    restart: unless-stopped
    ports:
      - "11434:11434"  # Exposed for local testing
    volumes:
      - ollama_data_local:/root/.ollama
    networks:
      - munich-news-network
    dns:
      - 8.8.8.8
      - 1.1.1.1
    # GPU support (uncomment if you have NVIDIA GPU)
    # deploy:
    #   resources:
    #     reservations:
    #       devices:
    #         - driver: nvidia
    #           count: all
    #           capabilities: [gpu]
    healthcheck:
      test: ["CMD-SHELL", "ollama list || exit 1"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 30s
  # Ollama Model Loader - Pulls phi3:latest (smaller model for local dev)
  ollama-setup:
    image: curlimages/curl:latest
    container_name: munich-news-local-ollama-setup
    depends_on:
      ollama:
        condition: service_healthy
    networks:
      - munich-news-network
    env_file:
      - backend/.env.local
    volumes:
      - ./scripts/setup-ollama-model.sh:/setup-ollama-model.sh:ro
    dns:
      - 8.8.8.8
      - 1.1.1.1
    command: sh /setup-ollama-model.sh
    restart: on-failure
  # Redis - Message queue for async tasks (Internal only - not exposed to host)
  redis:
    image: redis:7-alpine
    container_name: munich-news-local-redis
    restart: unless-stopped
    # No ports exposed - only accessible within Docker network
    networks:
      - munich-news-network
    healthcheck:
      test: ["CMD", "redis-cli", "ping"]
      interval: 30s
      timeout: 10s
      retries: 3
  # MongoDB Database (Exposed for local debugging)
  mongodb:
    image: mongo:latest
    container_name: munich-news-local-mongodb
    restart: unless-stopped
    ports:
      - "27017:27017"  # Exposed for local debugging
    environment:
      # For production, set MONGO_PASSWORD environment variable
      MONGO_INITDB_ROOT_USERNAME: ${MONGO_USERNAME:-admin}
      MONGO_INITDB_ROOT_PASSWORD: ${MONGO_PASSWORD:-changeme}
      MONGO_INITDB_DATABASE: munich_news
    volumes:
      - mongodb_data_local:/data/db
      - mongodb_config_local:/data/configdb
    networks:
      - munich-news-network
    command: mongod --bind_ip_all ${MONGO_AUTH:---auth}
    healthcheck:
      test: echo 'db.runCommand("ping").ok' | mongosh localhost:27017/test --quiet
      interval: 30s
      timeout: 10s
      retries: 3
  # News Crawler - Runs at 6 AM Berlin time
  crawler:
    build:
      context: .
      dockerfile: news_crawler/Dockerfile
    container_name: munich-news-local-crawler
    restart: unless-stopped
    depends_on:
      - mongodb
      - ollama
      - redis
    environment:
      - MONGODB_URI=mongodb://${MONGO_USERNAME:-admin}:${MONGO_PASSWORD:-changeme}@mongodb:27017/
      - REDIS_URL=redis://redis:6379
      - TZ=Europe/Berlin
    volumes:
      - ./backend/.env.local:/app/.env:ro
    networks:
      - munich-news-network
    healthcheck:
      test: ["CMD", "python", "-c", "import sys; sys.exit(0)"]
      interval: 1m
      timeout: 10s
      retries: 3
  # Backend API - Tracking and analytics
  backend:
    build:
      context: ./backend
      dockerfile: Dockerfile
    container_name: munich-news-local-backend
    restart: unless-stopped
    depends_on:
      - mongodb
      - redis
    ports:
      - "5001:5001"
    environment:
      - MONGODB_URI=mongodb://${MONGO_USERNAME:-admin}:${MONGO_PASSWORD:-changeme}@mongodb:27017/
      - REDIS_URL=redis://redis:6379
      - FLASK_PORT=5001
      - TZ=Europe/Berlin
    volumes:
      - ./backend/.env.local:/app/.env:ro
      - /var/run/docker.sock:/var/run/docker.sock:ro
    networks:
      - munich-news-network
    healthcheck:
      test: ["CMD", "python", "-c", "import urllib.request; urllib.request.urlopen('http://localhost:5001/health')"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 40s
  # Transport Crawler - API service for MVG disruptions (Internal only - not exposed to host)
  transport-crawler:
    build:
      context: ./transport_crawler
      dockerfile: Dockerfile
    container_name: munich-news-local-transport-crawler
    restart: unless-stopped
    depends_on:
      - mongodb
      - redis
    # No ports exposed - only accessible within Docker network
    environment:
      - MONGODB_URI=mongodb://${MONGO_USERNAME:-admin}:${MONGO_PASSWORD:-changeme}@mongodb:27017/
      - REDIS_URL=redis://redis:6379
      - TZ=Europe/Berlin
    volumes:
      - ./backend/.env.local:/app/.env:ro
    networks:
      - munich-news-network
    healthcheck:
      test: ["CMD", "python", "-c", "import urllib.request; urllib.request.urlopen('http://localhost:5002/health')"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 40s
  # Newsletter Sender - Runs at 7 AM Berlin time
  sender:
    build:
      context: .
      dockerfile: news_sender/Dockerfile
    container_name: munich-news-local-sender
    restart: unless-stopped
    depends_on:
      - mongodb
      - backend
      - crawler
      - transport-crawler
    environment:
      - MONGODB_URI=mongodb://${MONGO_USERNAME:-admin}:${MONGO_PASSWORD:-changeme}@mongodb:27017/
      - TZ=Europe/Berlin
    volumes:
      - ./backend/.env.local:/app/.env:ro
    networks:
      - munich-news-network
    healthcheck:
      test: ["CMD", "python", "-c", "import sys; sys.exit(0)"]
      interval: 1m
      timeout: 10s
      retries: 3
  # Frontend Web Interface
  frontend:
    build: ./frontend
    container_name: munich-news-local-frontend
    restart: unless-stopped
    # ports:
    #   - "3000:3000"
    ports:
      - "3000:3000"
    environment:
      - API_URL=http://backend:5001
      - PORT=3000
    depends_on:
      - backend
    networks:
      - munich-news-network
    healthcheck:
      test: ["CMD", "wget", "--quiet", "--tries=1", "--spider", "http://localhost:3000"]
      interval: 30s
      timeout: 10s
      retries: 3
 volumes:
  mongodb_data_local:
    driver: local
  mongodb_config_local:
    driver: local
  ollama_data_local:
    driver: local
 networks:
  munich-news-network:
    internal: false
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -1,20 +1,3 @@
 # Munich News Daily - Docker Compose Configuration
 #
 # GPU Support:
 #   To enable GPU acceleration for Ollama (5-10x faster):
 #   1. Check GPU availability: ./check-gpu.sh
 #   2. Start with GPU: ./start-with-gpu.sh
 #   Or manually: docker-compose -f docker-compose.yml -f docker-compose.gpu.yml up -d
 #
 # Security:
 #   - Only Backend API (port 5001) is exposed to host
 #   - MongoDB is internal-only (not exposed to host)
 #   - Ollama is internal-only (not exposed to host)
 #   - Crawler and Sender are internal-only
 #   All services communicate via internal Docker network
 #
 # See docs/OLLAMA_SETUP.md for detailed setup instructions
 services:
  # Ollama AI Service (Internal only - not exposed to host)
  ollama:
@@ -26,16 +9,11 @@ services:
      - ollama_data:/root/.ollama
    networks:
      - munich-news-network
-    # GPU support (uncomment if you have NVIDIA GPU)
+    dns:
-    # deploy:
+      - 8.8.8.8
-    #   resources:
+      - 1.1.1.1
    #     reservations:
    #       devices:
    #         - driver: nvidia
    #           count: all
    #           capabilities: [gpu]
    healthcheck:
-      test: ["CMD-SHELL", "ollama list || exit 1"]
+      test: [ "CMD-SHELL", "ollama list || exit 1" ]
      interval: 30s
      timeout: 10s
      retries: 3
@@ -54,6 +32,9 @@ services:
      - backend/.env
    volumes:
      - ./scripts/setup-ollama-model.sh:/setup-ollama-model.sh:ro
    dns:
      - 8.8.8.8
      - 1.1.1.1
    command: sh /setup-ollama-model.sh
    restart: on-failure
@@ -66,7 +47,7 @@ services:
    networks:
      - munich-news-network
    healthcheck:
-      test: ["CMD", "redis-cli", "ping"]
+      test: [ "CMD", "redis-cli", "ping" ]
      interval: 30s
      timeout: 10s
      retries: 3
@@ -94,6 +75,24 @@ services:
      timeout: 10s
      retries: 3
  # ChromaDB - Vector Database for AI features
  chromadb:
    image: chromadb/chroma:latest
    container_name: munich-news-chromadb
    restart: unless-stopped
    # No ports exposed - only accessible within Docker network
    environment:
      - IS_PERSISTENT=TRUE
    volumes:
      - chromadb_data:/chroma/chroma
    networks:
      - munich-news-network
    healthcheck:
      test: [ "CMD", "curl", "-f", "http://localhost:8000/api/v1/heartbeat" ]
      interval: 30s
      timeout: 10s
      retries: 3
  # News Crawler - Runs at 6 AM Berlin time
  crawler:
    build:
@@ -114,7 +113,7 @@ services:
    networks:
      - munich-news-network
    healthcheck:
-      test: ["CMD", "python", "-c", "import sys; sys.exit(0)"]
+      test: [ "CMD", "python", "-c", "import sys; sys.exit(0)" ]
      interval: 1m
      timeout: 10s
      retries: 3
@@ -143,7 +142,7 @@ services:
      - munich-news-network
      - proxy
    healthcheck:
-      test: ["CMD", "python", "-c", "import urllib.request; urllib.request.urlopen('http://localhost:5001/health')"]
+      test: [ "CMD", "python", "-c", "import urllib.request; urllib.request.urlopen('http://localhost:5001/health')" ]
      interval: 30s
      timeout: 10s
      retries: 3
@@ -182,7 +181,7 @@ services:
    networks:
      - munich-news-network
    healthcheck:
-      test: ["CMD", "python", "-c", "import urllib.request; urllib.request.urlopen('http://localhost:5002/health')"]
+      test: [ "CMD", "python", "-c", "import urllib.request; urllib.request.urlopen('http://localhost:5002/health')" ]
      interval: 30s
      timeout: 10s
      retries: 3
@@ -208,7 +207,7 @@ services:
    networks:
      - munich-news-network
    healthcheck:
-      test: ["CMD", "python", "-c", "import sys; sys.exit(0)"]
+      test: [ "CMD", "python", "-c", "import sys; sys.exit(0)" ]
      interval: 1m
      timeout: 10s
      retries: 3
@@ -233,7 +232,7 @@ services:
      - munich-news-network
      - proxy
    healthcheck:
-      test: ["CMD", "wget", "--quiet", "--tries=1", "--spider", "http://localhost:3000"]
+      test: [ "CMD", "wget", "--quiet", "--tries=1", "--spider", "http://localhost:3000" ]
      interval: 30s
      timeout: 10s
      retries: 3
@@ -258,9 +257,11 @@ volumes:
    driver: local
  ollama_data:
    driver: local
  chromadb_data:
    driver: local
 networks:
  munich-news-network:
-    internal: true
+    internal: false
  proxy:
    external: true
--- a/docs/LOCAL_DEVELOPMENT.md
+++ b/docs/LOCAL_DEVELOPMENT.md
@@ -0,0 +1,167 @@
 # Local Development Setup
 This guide helps you run Munich News Daily locally for development and testing.
 ## Quick Start
 ```bash
 # 1. Copy local environment files
 cp .env.local .env
 cp backend/.env.local backend/.env
 # 2. Start services with local configuration
 docker-compose -f docker-compose.local.yml up -d
 # 3. Check logs
 docker-compose -f docker-compose.local.yml logs -f
 # 4. Access services
 # - Frontend: http://localhost:3000
 # - Backend API: http://localhost:5001
 # - MongoDB: localhost:27017
 # - Ollama: http://localhost:11434
 ```
 ## Differences from Production
 | Feature | Production | Local Development |
 |---------|-----------|-------------------|
 | Ollama Model | `gemma3:12b` (large) | `phi3:latest` (small, fast) |
 | MongoDB Port | Internal only | Exposed on 27017 |
 | Ollama Port | Internal only | Exposed on 11434 |
 | Container Names | `munich-news-*` | `munich-news-local-*` |
 | Volumes | `*_data` | `*_data_local` |
 | Email | Production SMTP | Test/disabled |
 ## Useful Commands
 ### Start/Stop Services
 ```bash
 # Start all services
 docker-compose -f docker-compose.local.yml up -d
 # Stop all services
 docker-compose -f docker-compose.local.yml down
 # Restart a specific service
 docker-compose -f docker-compose.local.yml restart backend
 # View logs
 docker-compose -f docker-compose.local.yml logs -f crawler
 ```
 ### Testing
 ```bash
 # Trigger a news crawl (2 articles for quick testing)
 curl -X POST http://localhost:5001/api/admin/trigger-crawl \
  -H "Content-Type: application/json" \
  -d '{"max_articles": 2}'
 # Trigger transport crawl
 curl -X POST http://localhost:5001/api/transport/crawl
 # Check articles in MongoDB
 docker exec munich-news-local-mongodb mongosh munich_news \
  --eval "db.articles.find({}, {title: 1, keywords: 1, category: 1}).limit(3)"
 # Check transport disruptions
 curl http://localhost:5001/api/transport/disruptions
 ```
 ### Database Access
 ```bash
 # Connect to MongoDB
 docker exec -it munich-news-local-mongodb mongosh munich_news
 # Or from host (if you have mongosh installed)
 mongosh "mongodb://admin:local123@localhost:27017/munich_news"
 # Useful queries
 db.articles.countDocuments()
 db.articles.find({keywords: {$exists: true}}).limit(5)
 db.subscribers.find()
 db.transport_alerts.find()
 ```
 ### Ollama Testing
 ```bash
 # List models
 curl http://localhost:11434/api/tags
 # Test generation
 curl http://localhost:11434/api/generate -d '{
  "model": "phi3:latest",
  "prompt": "Summarize: Munich opens new U-Bahn line",
  "stream": false
 }'
 ```
 ## Cleanup
 ```bash
 # Stop and remove containers
 docker-compose -f docker-compose.local.yml down
 # Remove volumes (WARNING: deletes all data)
 docker-compose -f docker-compose.local.yml down -v
 # Remove local volumes specifically
 docker volume rm munich-news_mongodb_data_local
 docker volume rm munich-news_mongodb_config_local
 docker volume rm munich-news_ollama_data_local
 ```
 ## Switching Between Local and Production
 ```bash
 # Switch to local
 cp .env.local .env
 cp backend/.env.local backend/.env
 docker-compose -f docker-compose.local.yml up -d
 # Switch to production
 cp .env.production .env  # (if you have one)
 cp backend/.env.production backend/.env
 docker-compose up -d
 ```
 ## Troubleshooting
 ### Ollama model not downloading
 ```bash
 # Pull model manually
 docker exec munich-news-local-ollama ollama pull phi3:latest
 ```
 ### MongoDB connection refused
 ```bash
 # Check if MongoDB is running
 docker-compose -f docker-compose.local.yml ps mongodb
 # Check logs
 docker-compose -f docker-compose.local.yml logs mongodb
 ```
 ### Port already in use
 ```bash
 # Check what's using the port
 lsof -i :5001  # or :3000, :27017, etc.
 # Stop the conflicting service or change port in docker-compose.local.yml
 ```
 ## Tips
 1. **Use phi3 for speed** - It's much faster than gemma3 for local testing
 2. **Limit articles** - Use `max_articles: 2` for quick crawl tests
 3. **Watch logs** - Keep logs open to see what's happening
 4. **Separate volumes** - Local and production use different volumes, so they don't interfere
 ## Next Steps
 - See `docs/PERSONALIZATION.md` for personalization feature development
 - See `docs/OLLAMA_SETUP.md` for AI configuration
 - See main `README.md` for general documentation
--- a/docs/PERSONALIZATION.md
+++ b/docs/PERSONALIZATION.md
@@ -0,0 +1,217 @@
 # Newsletter Personalization Implementation
 ## Overview
 Personalized newsletters based on user click behavior, using keywords and categories to build interest profiles.
 ## Implementation Phases
 ### ✅ Phase 1: Keyword Extraction (COMPLETED)
 **Status:** Implemented
 **Files Modified:**
 - `news_crawler/ollama_client.py` - Added `extract_keywords()` method
 - `news_crawler/crawler_service.py` - Integrated keyword extraction into crawl process
 **What it does:**
 - Extracts 5 keywords from each article using Ollama AI
 - Keywords stored in `articles` collection: `keywords: ["Bayern Munich", "Football", ...]`
 - Runs automatically during news crawling
 **Test it:**
 ```bash
 # Trigger a crawl
 curl -X POST http://localhost:5001/api/admin/trigger-crawl -d '{"max_articles": 2}'
 # Check articles have keywords
 docker exec munich-news-mongodb mongosh munich_news --eval "db.articles.findOne({}, {title: 1, keywords: 1})"
 ```
 ---
 ### ✅ Phase 2: Click Tracking Enhancement (COMPLETED)
 **Status:** Implemented
 **Goal:** Track clicks with keyword metadata
 **Files Modified:**
 - `backend/services/tracking_service.py` - Enhanced `create_newsletter_tracking()` to look up article metadata
 **What it does:**
 - When creating tracking links, looks up article from database
 - Stores article ID, category, and keywords in tracking record
 - Enables building user interest profiles from click behavior
 **Database Schema:**
 ```javascript
 // link_clicks collection
 {
  tracking_id: "uuid",
  newsletter_id: "2024-11-18",
  subscriber_email: "user@example.com",
  article_url: "https://...",
  article_title: "Article Title",
  article_id: "673abc123...",              // NEW: Article database ID
  category: "sports",                      // NEW: Article category
  keywords: ["Bayern Munich", "Bundesliga"], // NEW: Keywords for personalization
  clicked: false,
  clicked_at: null,
  user_agent: null,
  created_at: ISODate()
 }
 ```
 **Test it:**
 ```bash
 # Send a test newsletter
 curl -X POST http://localhost:5001/api/admin/send-newsletter
 # Check tracking records have keywords
 docker exec munich-news-mongodb mongosh munich_news --eval "db.link_clicks.findOne({}, {article_title: 1, keywords: 1, category: 1})"
 ```
 ---
 ### ✅ Phase 3: User Interest Profiling (COMPLETED)
 **Status:** Implemented
 **Goal:** Build user interest profiles from click history
 **Files Created:**
 - `backend/services/interest_profiling_service.py` - Core profiling logic
 - `backend/routes/interests_routes.py` - API endpoints for interest management
 **Files Modified:**
 - `backend/routes/tracking_routes.py` - Auto-update interests on click
 - `backend/app.py` - Register interests routes
 **What it does:**
 - Automatically builds interest profiles when users click articles
 - Tracks interest scores for categories and keywords (0.0 to 1.0)
 - Increments scores by 0.1 per click, capped at 1.0
 - Provides decay mechanism for old interests
 - Supports rebuilding profiles from click history
 **Database Schema:**
 ```javascript
 // user_interests collection
 {
  email: "user@example.com",
  categories: {
    sports: 0.8,
    local: 0.5,
    science: 0.2
  },
  keywords: {
    "Bayern Munich": 0.9,
    "Oktoberfest": 0.7,
    "AI": 0.3
  },
  total_clicks: 15,
  last_updated: ISODate(),
  created_at: ISODate()
 }
 ```
 **API Endpoints:**
 ```bash
 # Get user interests
 GET /api/interests/<email>
 # Get top interests
 GET /api/interests/<email>/top?top_n=10
 # Rebuild from history
 POST /api/interests/<email>/rebuild
 Body: {"days_lookback": 30}
 # Decay old interests
 POST /api/interests/decay
 Body: {"decay_factor": 0.95, "days_threshold": 7}
 # Get statistics
 GET /api/interests/statistics
 # Delete profile (GDPR)
 DELETE /api/interests/<email>
 ```
 **Test it:**
 ```bash
 # Run test script
 docker exec munich-news-local-backend python test_interest_profiling.py
 # View a user's interests
 curl http://localhost:5001/api/interests/user@example.com
 # Get statistics
 curl http://localhost:5001/api/interests/statistics
 ```
 ---
 ### ✅ Phase 4: Personalized Newsletter (COMPLETED)
 **Status:** Implemented
 **Goal:** Rank and select articles based on user interests
 **Files Created:**
 - `backend/services/personalization_service.py` - Core personalization logic
 - `backend/routes/personalization_routes.py` - API endpoints for testing
 **Files Modified:**
 - `backend/app.py` - Register personalization routes
 **What it does:**
 - Scores articles based on user's category and keyword interests
 - Ranks articles by personalization score (0.0 to 1.0)
 - Selects mix of personalized (70%) + trending (30%) content
 - Provides explanations for recommendations
 **Algorithm:**
 ```python
 score = (category_match * 0.4) + (keyword_match * 0.6)
 # Example:
 # User interests: sports=0.8, "Bayern Munich"=0.9
 # Article: sports category, keywords=["Bayern Munich", "Football"]
 # Score = (0.8 * 0.4) + (0.9 * 0.6) = 0.32 + 0.54 = 0.86
 ```
 **API Endpoints:**
 ```bash
 # Preview personalized newsletter
 GET /api/personalize/preview/<email>?max_articles=10&hours_lookback=24
 # Explain recommendation
 POST /api/personalize/explain
 Body: {"email": "user@example.com", "article_id": "..."}
 ```
 **Test it:**
 ```bash
 # Run test script
 docker exec munich-news-local-backend python test_personalization.py
 # Preview personalized newsletter
 curl "http://localhost:5001/api/personalize/preview/demo@example.com?max_articles=5"
 ```
 ---
 ## ✅ All Phases Complete!
 1. ~~**Phase 1:** Keyword extraction from articles~~ ✅ DONE
 2. ~~**Phase 2:** Click tracking with keywords~~ ✅ DONE
 3. ~~**Phase 3:** User interest profiling~~ ✅ DONE
 4. ~~**Phase 4:** Personalized newsletter generation~~ ✅ DONE
 ## Next Steps for Production
 1. **Integrate with newsletter sender** - Modify `news_sender/sender_service.py` to use personalization
 2. **A/B testing** - Compare personalized vs non-personalized engagement
 3. **Tune parameters** - Adjust personalization_ratio, weights, decay rates
 4. **Monitor metrics** - Track click-through rates, open rates by personalization score
 5. **User controls** - Add UI for users to view/edit their interests
 ## Configuration
 No configuration needed yet. Keyword extraction uses existing Ollama settings from `backend/.env`:
 - `OLLAMA_ENABLED=true`
 - `OLLAMA_MODEL=gemma3:12b`
 - `OLLAMA_BASE_URL=http://ollama:11434`
--- a/docs/PERSONALIZATION_COMPLETE.md
+++ b/docs/PERSONALIZATION_COMPLETE.md
@@ -0,0 +1,195 @@
 # 🎉 Newsletter Personalization System - Complete!
 All 4 phases of the personalization system have been successfully implemented and tested.
 ## ✅ What Was Built
 ### Phase 1: Keyword Extraction
 - AI-powered keyword extraction from articles using Ollama
 - 5 keywords per article automatically extracted during crawling
 - Keywords stored in database for personalization
 ### Phase 2: Click Tracking Enhancement  
 - Enhanced tracking to capture article keywords and category
 - Tracking records now include metadata for building interest profiles
 - Privacy-compliant with opt-out and GDPR support
 ### Phase 3: User Interest Profiling
 - Automatic profile building from click behavior
 - Interest scores (0.0-1.0) for categories and keywords
 - Decay mechanism for old interests
 - API endpoints for viewing and managing profiles
 ### Phase 4: Personalized Newsletter Generation
 - Article scoring based on user interests
 - Smart ranking algorithm (40% category + 60% keywords)
 - Mix of personalized (70%) + trending (30%) content
 - Explanation system for recommendations
 ## 📊 How It Works
 ```
 1. User clicks article in newsletter
   ↓
 2. System records: keywords + category
   ↓
 3. Interest profile updates automatically
   ↓
 4. Next newsletter: articles ranked by interests
   ↓
 5. User receives personalized content
 ```
 ## 🧪 Testing
 All phases have been tested and verified:
 ```bash
 # Run comprehensive test suite (tests all 4 phases)
 docker exec munich-news-local-backend python test_personalization_system.py
 # Or test keyword extraction separately
 docker exec munich-news-local-crawler python -c "from crawler_service import crawl_all_feeds; crawl_all_feeds(max_articles_per_feed=2)"
 ```
 ## 🔌 API Endpoints
 ### Interest Management
 ```bash
 GET    /api/interests/<email>              # View profile
 GET    /api/interests/<email>/top          # Top interests
 POST   /api/interests/<email>/rebuild      # Rebuild from history
 GET    /api/interests/statistics           # Platform stats
 DELETE /api/interests/<email>              # Delete (GDPR)
 ```
 ### Personalization
 ```bash
 GET  /api/personalize/preview/<email>      # Preview personalized newsletter
 POST /api/personalize/explain              # Explain recommendation
 ```
 ## 📈 Example Results
 ### User Profile
 ```json
 {
  "email": "user@example.com",
  "categories": {
    "sports": 0.30,
    "local": 0.10
  },
  "keywords": {
    "Bayern Munich": 0.30,
    "Football": 0.20,
    "Transportation": 0.10
  },
  "total_clicks": 5
 }
 ```
 ### Personalized Newsletter
 ```json
 {
  "articles": [
    {
      "title": "Bayern Munich wins championship",
      "personalization_score": 0.86,
      "category": "sports",
      "keywords": ["Bayern Munich", "Football"]
    },
    {
      "title": "New S-Bahn line opens",
      "personalization_score": 0.42,
      "category": "local",
      "keywords": ["Transportation", "Munich"]
    }
  ],
  "statistics": {
    "highly_personalized": 1,
    "moderately_personalized": 1,
    "trending": 0
  }
 }
 ```
 ## 🎯 Scoring Algorithm
 ```python
 # Article score calculation
 category_score = user_interests.categories[article.category]
 keyword_score = average(user_interests.keywords[kw] for kw in article.keywords)
 final_score = (category_score * 0.4) + (keyword_score * 0.6)
 ```
 **Example:**
 - User: sports=0.8, "Bayern Munich"=0.9
 - Article: sports category, keywords=["Bayern Munich", "Football"]
 - Score = (0.8 × 0.4) + (0.9 × 0.6) = 0.32 + 0.54 = **0.86**
 ## 🚀 Production Integration
 To integrate with the newsletter sender:
 1. **Modify `news_sender/sender_service.py`:**
 ```python
 from services.personalization_service import select_personalized_articles
 # For each subscriber
 personalized_articles = select_personalized_articles(
    all_articles,
    subscriber_email,
    max_articles=10
 )
 ```
 2. **Enable personalization flag in config:**
 ```env
 PERSONALIZATION_ENABLED=true
 PERSONALIZATION_RATIO=0.7  # 70% personalized, 30% trending
 ```
 3. **Monitor metrics:**
 - Click-through rate by personalization score
 - Open rates for personalized vs non-personalized
 - User engagement over time
 ## 🔐 Privacy & Compliance
 - ✅ Users can opt out of tracking
 - ✅ Interest profiles can be deleted (GDPR)
 - ✅ Automatic anonymization after 90 days
 - ✅ No PII beyond email address
 - ✅ Transparent recommendation explanations
 ## 📁 Files Created/Modified
 ### New Files
 - `backend/services/interest_profiling_service.py`
 - `backend/services/personalization_service.py`
 - `backend/routes/interests_routes.py`
 - `backend/routes/personalization_routes.py`
 - `backend/test_tracking_phase2.py`
 - `backend/test_interest_profiling.py`
 - `backend/test_personalization.py`
 - `docs/PERSONALIZATION.md`
 ### Modified Files
 - `news_crawler/ollama_client.py` - Added keyword extraction
 - `news_crawler/crawler_service.py` - Integrated keyword extraction
 - `backend/services/tracking_service.py` - Enhanced with metadata
 - `backend/routes/tracking_routes.py` - Auto-update interests
 - `backend/app.py` - Registered new routes
 ## 🎓 Key Learnings
 1. **Incremental scoring works well** - 0.1 per click prevents over-weighting
 2. **Mix is important** - 70/30 personalized/trending avoids filter bubbles
 3. **Keywords > Categories** - 60/40 weight reflects keyword importance
 4. **Decay is essential** - Prevents stale interests from dominating
 5. **Transparency matters** - Explanation API helps users understand recommendations
 ## 🎉 Status: COMPLETE
 All 4 phases implemented, tested, and documented. The personalization system is ready for production integration!
--- a/frontend/public/admin.js
+++ b/frontend/public/admin.js
@@ -27,7 +27,16 @@ function refreshAll() {
 // Load system statistics
 async function loadSystemStats() {
    try {
-        const response = await fetch(`${API_BASE}/api/stats`);
+        const controller = new AbortController();
        const timeoutId = setTimeout(() => controller.abort(), 10000); // 10 second timeout
        const response = await fetch(`${API_BASE}/api/stats`, { signal: controller.signal });
        clearTimeout(timeoutId);
        if (!response.ok) {
            throw new Error(`HTTP ${response.status}: ${response.statusText}`);
        }
        const data = await response.json();
        const html = `
@@ -59,14 +68,25 @@ async function loadSystemStats() {
        document.getElementById('systemStats').innerHTML = html;
    } catch (error) {
-        document.getElementById('systemStats').innerHTML = `<div class="error">Error loading stats: ${error.message}</div>`;
+        console.error('Error loading system stats:', error);
        const errorMsg = error.name === 'AbortError' ? 'Request timeout' : error.message;
        document.getElementById('systemStats').innerHTML = `<div class="error">Error: ${errorMsg}</div>`;
    }
 }
 // Load Ollama status
 async function loadOllamaStatus() {
    try {
-        const response = await fetch(`${API_BASE}/api/ollama/ping`);
+        const controller = new AbortController();
        const timeoutId = setTimeout(() => controller.abort(), 10000);
        const response = await fetch(`${API_BASE}/api/ollama/ping`, { signal: controller.signal });
        clearTimeout(timeoutId);
        if (!response.ok) {
            throw new Error(`HTTP ${response.status}`);
        }
        const data = await response.json();
        const isActive = data.status === 'success';
@@ -102,25 +122,37 @@ async function loadOllamaStatus() {
        document.getElementById('ollamaStatus').innerHTML = html;
    } catch (error) {
-        document.getElementById('ollamaStatus').innerHTML = `<div class="error">Error: ${error.message}</div>`;
+        console.error('Error loading Ollama status:', error);
        const errorMsg = error.name === 'AbortError' ? 'Request timeout' : error.message;
        document.getElementById('ollamaStatus').innerHTML = `<div class="error">Error: ${errorMsg}</div>`;
    }
 }
 // Load GPU status
 async function loadGPUStatus() {
    try {
-        const response = await fetch(`${API_BASE}/api/ollama/gpu-status`);
+        const controller = new AbortController();
        const timeoutId = setTimeout(() => controller.abort(), 10000);
        const response = await fetch(`${API_BASE}/api/ollama/gpu-status`, { signal: controller.signal });
        clearTimeout(timeoutId);
        if (!response.ok) {
            throw new Error(`HTTP ${response.status}`);
        }
        const data = await response.json();
-        const gpuActive = data.gpu_in_use;
+        const gpuActive = data.gpu_in_use || false;
        const gpuAvailable = data.gpu_available || false;
        const statusClass = gpuActive ? 'status-active' : 'status-warning';
        const html = `
            <div class="stat-row">
                <span class="stat-label">GPU Available</span>
                <span class="stat-value">
-                    <span class="status-indicator ${data.gpu_available ? 'status-active' : 'status-inactive'}"></span>
+                    <span class="status-indicator ${gpuAvailable ? 'status-active' : 'status-inactive'}"></span>
-                    ${data.gpu_available ? 'Yes' : 'No'}
+                    ${gpuAvailable ? 'Yes' : 'No'}
                </span>
            </div>
            <div class="stat-row">
@@ -134,16 +166,6 @@ async function loadGPUStatus() {
                <span class="stat-label">Models Loaded</span>
                <span class="stat-value">${data.models_loaded || 0}</span>
            </div>
            ${data.gpu_details ? `
            <div class="stat-row">
                <span class="stat-label">GPU Model</span>
                <span class="stat-value">${data.gpu_details.model}</span>
            </div>
            <div class="stat-row">
                <span class="stat-label">GPU Layers</span>
                <span class="stat-value">${data.gpu_details.gpu_layers}</span>
            </div>
            ` : ''}
            ${!gpuActive ? `
            <div style="margin-top: 10px; padding: 10px; background: #fef3c7; border-radius: 5px; font-size: 12px;">
                💡 Enable GPU for 5-10x faster processing
@@ -153,7 +175,9 @@ async function loadGPUStatus() {
        document.getElementById('gpuStatus').innerHTML = html;
    } catch (error) {
-        document.getElementById('gpuStatus').innerHTML = `<div class="error">Error: ${error.message}</div>`;
+        console.error('Error loading GPU status:', error);
        const errorMsg = error.name === 'AbortError' ? 'Request timeout' : error.message;
        document.getElementById('gpuStatus').innerHTML = `<div class="error">Error: ${errorMsg}</div>`;
    }
 }
@@ -204,7 +228,16 @@ async function runPerformanceTest() {
 // Load available models
 async function loadModels() {
    try {
-        const response = await fetch(`${API_BASE}/api/ollama/models`);
+        const controller = new AbortController();
        const timeoutId = setTimeout(() => controller.abort(), 10000);
        const response = await fetch(`${API_BASE}/api/ollama/models`, { signal: controller.signal });
        clearTimeout(timeoutId);
        if (!response.ok) {
            throw new Error(`HTTP ${response.status}`);
        }
        const data = await response.json();
        if (data.models && data.models.length > 0) {
@@ -227,14 +260,25 @@ async function loadModels() {
            document.getElementById('modelsList').innerHTML = '<div>No models found</div>';
        }
    } catch (error) {
-        document.getElementById('modelsList').innerHTML = `<div class="error">Error: ${error.message}</div>`;
+        console.error('Error loading models:', error);
        const errorMsg = error.name === 'AbortError' ? 'Request timeout' : error.message;
        document.getElementById('modelsList').innerHTML = `<div class="error">Error: ${errorMsg}</div>`;
    }
 }
 // Load configuration
 async function loadConfig() {
    try {
-        const response = await fetch(`${API_BASE}/api/ollama/config`);
+        const controller = new AbortController();
        const timeoutId = setTimeout(() => controller.abort(), 10000);
        const response = await fetch(`${API_BASE}/api/ollama/config`, { signal: controller.signal });
        clearTimeout(timeoutId);
        if (!response.ok) {
            throw new Error(`HTTP ${response.status}`);
        }
        const data = await response.json();
        const html = `
@@ -262,14 +306,25 @@ async function loadConfig() {
        document.getElementById('configInfo').innerHTML = html;
    } catch (error) {
-        document.getElementById('configInfo').innerHTML = `<div class="error">Error: ${error.message}</div>`;
+        console.error('Error loading config:', error);
        const errorMsg = error.name === 'AbortError' ? 'Request timeout' : error.message;
        document.getElementById('configInfo').innerHTML = `<div class="error">Error: ${errorMsg}</div>`;
    }
 }
 // Load clustering statistics
 async function loadClusteringStats() {
    try {
-        const response = await fetch(`${API_BASE}/api/stats`);
+        const controller = new AbortController();
        const timeoutId = setTimeout(() => controller.abort(), 10000);
        const response = await fetch(`${API_BASE}/api/stats`, { signal: controller.signal });
        clearTimeout(timeoutId);
        if (!response.ok) {
            throw new Error(`HTTP ${response.status}`);
        }
        const data = await response.json();
        const clusteringRate = data.clustered_articles > 0 
@@ -310,7 +365,9 @@ async function loadClusteringStats() {
        document.getElementById('clusteringStats').innerHTML = html;
    } catch (error) {
-        document.getElementById('clusteringStats').innerHTML = `<div class="error">Error: ${error.message}</div>`;
+        console.error('Error loading clustering stats:', error);
        const errorMsg = error.name === 'AbortError' ? 'Request timeout' : error.message;
        document.getElementById('clusteringStats').innerHTML = `<div class="error">Error: ${errorMsg}</div>`;
    }
 }
@@ -512,7 +569,16 @@ async function loadRecentArticles() {
    const container = document.getElementById('recentArticles');
    try {
-        const response = await fetch('/api/admin/recent-articles');
+        const controller = new AbortController();
        const timeoutId = setTimeout(() => controller.abort(), 10000);
        const response = await fetch('/api/admin/recent-articles', { signal: controller.signal });
        clearTimeout(timeoutId);
        if (!response.ok) {
            throw new Error(`HTTP ${response.status}`);
        }
        const data = await response.json();
        if (data.articles && data.articles.length > 0) {
@@ -548,7 +614,9 @@ async function loadRecentArticles() {
            container.innerHTML = '<p style="color: #666;">No summarized articles found.</p>';
        }
    } catch (error) {
-        container.innerHTML = '<p style="color: red;">Failed to load recent articles</p>';
+        console.error('Error loading recent articles:', error);
        const errorMsg = error.name === 'AbortError' ? 'Request timeout' : error.message;
        container.innerHTML = `<p style="color: red;">Error: ${errorMsg}</p>`;
    }
 }
--- a/frontend/public/app.js
+++ b/frontend/public/app.js
@@ -113,10 +113,15 @@ function setupInfiniteScroll() {
 }
 // Search functionality
-function handleSearch() {
+let searchTimeout;
 async function handleSearch() {
    const searchInput = document.getElementById('searchInput');
    const clearBtn = document.getElementById('clearSearch');
-    searchQuery = searchInput.value.trim().toLowerCase();
+    const searchStats = document.getElementById('searchStats');
    const newsGrid = document.getElementById('newsGrid');
    searchQuery = searchInput.value.trim();
    // Show/hide clear button
    if (searchQuery) {
@@ -125,41 +130,68 @@ function handleSearch() {
        clearBtn.classList.add('hidden');
    }
-    // Filter articles
+    // Clear previous timeout
    if (searchTimeout) clearTimeout(searchTimeout);
    // If empty query, reset to all articles
    if (searchQuery === '') {
        filteredArticles = allArticles;
-    } else {
+        displayedCount = 0;
-        filteredArticles = allArticles.filter(article => {
+        newsGrid.innerHTML = '';
-            const title = article.title.toLowerCase();
+        updateSearchStats();
-            const summary = (article.summary || '').toLowerCase().replace(/<[^>]*>/g, '');
+        loadMoreArticles();
-            const source = formatSourceName(article.source).toLowerCase();
+        return;
            return title.includes(searchQuery) || 
                   summary.includes(searchQuery) || 
                   source.includes(searchQuery);
        });
    }
-    // Reset display
+    // Debounce search API call
    searchTimeout = setTimeout(async () => {
        // Show searching state
        newsGrid.innerHTML = '<div class="text-center py-10 text-gray-500">Searching...</div>';
        try {
            const response = await fetch(`/api/search?q=${encodeURIComponent(searchQuery)}&limit=20`);
            // Check if response is ok
            if (!response.ok) {
                const errorText = await response.text();
                throw new Error(`Server returned ${response.status}: ${errorText}`);
            }
            const data = await response.json();
            if (data.results && data.results.length > 0) {
                // Map results to match card format
                filteredArticles = data.results.map(item => ({
                    title: item.title,
                    link: item.link,
                    source: item.source,
                    summary: item.snippet, // Map snippet to summary
                    published_at: item.published_at,
                    score: item.relevance_score
                }));
                displayedCount = 0;
    const newsGrid = document.getElementById('newsGrid');
                newsGrid.innerHTML = '';
                // Update stats
-    updateSearchStats();
+                searchStats.textContent = `Found ${filteredArticles.length} relevant articles`;
    // Load filtered articles
    if (filteredArticles.length > 0) {
                loadMoreArticles();
            } else {
                newsGrid.innerHTML = `
                    <div class="text-center py-16">
                        <div class="text-6xl mb-4">🔍</div>
-                <p class="text-xl text-gray-600 mb-2">No articles found</p>
+                        <p class="text-xl text-gray-600 mb-2">No relevant articles found</p>
-                <p class="text-gray-400">Try a different search term</p>
+                        <p class="text-gray-400">Try different keywords or concepts</p>
                    </div>
                `;
                searchStats.textContent = 'No results found';
            }
        } catch (error) {
            console.error('Search failed:', error);
            newsGrid.innerHTML = `<div class="text-center py-10 text-red-400">Search failed: ${error.message}</div>`;
        }
    }, 500); // 500ms debounce
 }
 function clearSearch() {
@@ -423,7 +455,7 @@ async function unsubscribe() {
 }
 // Close modal when clicking outside
-window.onclick = function(event) {
+window.onclick = function (event) {
    const modal = document.getElementById('unsubscribeModal');
    if (event.target === modal) {
        closeUnsubscribe();
--- a/frontend/server.js
+++ b/frontend/server.js
@@ -70,6 +70,17 @@ app.post('/api/unsubscribe', async (req, res) => {
  }
 });
 app.get('/api/subscribers', async (req, res) => {
  try {
    const response = await axios.get(`${API_URL}/api/subscribers`);
    res.json(response.data);
  } catch (error) {
    res.status(error.response?.status || 500).json(
      error.response?.data || { error: 'Failed to get subscribers' }
    );
  }
 });
 app.get('/api/subscribers/:email', async (req, res) => {
  try {
    const response = await axios.get(`${API_URL}/api/subscribers/${req.params.email}`);
@@ -193,6 +204,31 @@ app.get('/api/ollama/config', async (req, res) => {
  }
 });
 app.get('/api/search', async (req, res) => {
  try {
    const { q, limit, category } = req.query;
    const response = await axios.get(`${API_URL}/api/search`, {
      params: { q, limit, category }
    });
    res.json(response.data);
  } catch (error) {
    if (error.response) {
      // The request was made and the server responded with a status code
      // that falls out of the range of 2xx
      console.error('Search API Error:', error.response.status, error.response.data);
      res.status(error.response.status).json(error.response.data);
    } else if (error.request) {
      // The request was made but no response was received
      console.error('Search API No Response:', error.request);
      res.status(502).json({ error: 'Search service unavailable (timeout/connection)' });
    } else {
      // Something happened in setting up the request that triggered an Error
      console.error('Search API Request Error:', error.message);
      res.status(500).json({ error: 'Internal proxy error' });
    }
  }
 });
 app.listen(PORT, () => {
  console.log(`Frontend server running on http://localhost:${PORT}`);
  console.log(`Admin dashboard: http://localhost:${PORT}/admin.html`);
--- a/news_crawler/chroma_client.py
+++ b/news_crawler/chroma_client.py
@@ -0,0 +1,166 @@
 """
 ChromaDB Client for storing and retrieving document embeddings
 """
 import chromadb
 from chromadb.config import Settings
 from chromadb.utils import embedding_functions
 import logging
 import os
 import time
 class ChromaClient:
    """
    Client for interacting with ChromaDB vector database.
    Uses Ollama for generating embeddings if available, otherwise falls back to default.
    """
    def __init__(self, host, port, collection_name='munich_news_articles', ollama_base_url=None):
        """
        Initialize ChromaDB client
        Args:
            host: ChromaDB host (e.g. 'localhost' or 'chromadb')
            port: ChromaDB port (default 8000)
            collection_name: Name of the collection to use
            ollama_base_url: Optional URL for Ollama embedding function
        """
        self.host = host
        self.port = port
        self.collection_name = collection_name
        self.client = None
        self.collection = None
        # Setup embedding function
        # We prefer using a local embedding model compatible with Ollama or SentenceTransformers
        # For simplicity in this stack, we can use the default SentenceTransformer (all-MiniLM-L6-v2)
        # which is downloaded automatically by chromadb utils.
        # Alternatively, we could define a custom function using Ollama's /api/embeddings
        self.embedding_function = embedding_functions.DefaultEmbeddingFunction()
    def connect(self):
        """Establish connection to ChromaDB"""
        try:
            self.client = chromadb.HttpClient(
                host=self.host,
                port=self.port,
                settings=Settings(allow_reset=True, anonymized_telemetry=False)
            )
            # Create or get collection
            self.collection = self.client.get_or_create_collection(
                name=self.collection_name,
                embedding_function=self.embedding_function,
                metadata={"hnsw:space": "cosine"}
            )
            print(f"✓ Connected to ChromaDB at {self.host}:{self.port}")
            return True
        except Exception as e:
            print(f"⚠ Could not connect to ChromaDB: {e}")
            return False
    def add_articles(self, articles):
        """
        Add articles to the vector database
        Args:
            articles: List of dictionaries containing article data.
                     Must have 'link' (used as ID), 'title', 'content', etc.
        """
        if not self.client or not self.collection:
            if not self.connect():
                return False
        if not articles:
            return True
        ids = []
        documents = []
        metadatas = []
        for article in articles:
            # Skip if critical data missing
            if not article.get('link') or not article.get('content'):
                continue
            # Use link as unique ID
            article_id = article.get('link')
            # Prepare text for embedding (Title + Summary + Start of Content)
            # This gives semantic search a good overview
            # Use English title if available, otherwise original
            title = article.get('title_en') if article.get('title_en') else article.get('title', '')
            summary = article.get('summary') or ''
            content_snippet = article.get('content', '')[:1000]
            text_to_embed = f"{title}\n\n{summary}\n\n{content_snippet}"
            # robust metadata (flat dict, no nested objects)
            metadata = {
                "title": title[:100], # Truncate for metadata limits
                "url": article_id,
                "source": article.get('source', 'unknown'),
                "category": article.get('category', 'general'),
                "published_at": str(article.get('published_at', '')),
                "mongo_id": str(article.get('_id', ''))
            }
            ids.append(article_id)
            documents.append(text_to_embed)
            metadatas.append(metadata)
        if not ids:
            return True
        try:
            self.collection.upsert(
                ids=ids,
                documents=documents,
                metadatas=metadatas
            )
            print(f"✓ Indexed {len(ids)} articles in ChromaDB")
            return True
        except Exception as e:
            print(f"✗ Failed to index in ChromaDB: {e}")
            return False
    def search(self, query_text, n_results=5, where=None):
        """
        Search for relevant articles
        Args:
            query_text: The search query
            n_results: Number of results to return
            where: Metadata filter dict (e.g. {"category": "sports"})
        """
        if not self.client or not self.collection:
            if not self.connect():
                return []
        try:
            results = self.collection.query(
                query_texts=[query_text],
                n_results=n_results,
                where=where
            )
            # Format results into a nice list of dicts
            formatted_results = []
            if results and results['ids']:
                for i, id in enumerate(results['ids'][0]):
                    item = {
                        'id': id,
                        'document': results['documents'][0][i] if results['documents'] else None,
                        'metadata': results['metadatas'][0][i] if results['metadatas'] else {},
                        'distance': results['distances'][0][i] if results['distances'] else 0
                    }
                    formatted_results.append(item)
            return formatted_results
        except Exception as e:
            print(f"✗ Search failed: {e}")
            return []
 if __name__ == "__main__":
    # Test client
    client = ChromaClient(host='localhost', port=8000)
    client.connect()
--- a/news_crawler/config.py
+++ b/news_crawler/config.py
@@ -34,6 +34,11 @@ class Config:
    MONGODB_URI = os.getenv('MONGODB_URI', 'mongodb://localhost:27017/')
    DB_NAME = 'munich_news'
    # ChromaDB Configuration
    CHROMA_HOST = os.getenv('CHROMA_HOST', 'chromadb')
    CHROMA_PORT = int(os.getenv('CHROMA_PORT', '8000'))
    CHROMA_COLLECTION = 'munich_news_articles'
    # Ollama Configuration
    OLLAMA_BASE_URL = os.getenv('OLLAMA_BASE_URL', 'http://localhost:11434')
    OLLAMA_MODEL = os.getenv('OLLAMA_MODEL', 'phi3:latest')
--- a/news_crawler/crawler_service.py
+++ b/news_crawler/crawler_service.py
@@ -14,7 +14,9 @@ from rss_utils import extract_article_url, extract_article_summary, extract_publ
 from config import Config
 from ollama_client import OllamaClient
 from article_clustering import ArticleClusterer
 from article_clustering import ArticleClusterer
 from cluster_summarizer import create_cluster_summaries
 from chroma_client import ChromaClient
 # Load environment variables
 load_dotenv(dotenv_path='../.env')
@@ -38,6 +40,13 @@ ollama_client = OllamaClient(
 # Initialize Article Clusterer (will be initialized after ollama_client)
 article_clusterer = None
 # Initialize ChromaDB client
 chroma_client = ChromaClient(
    host=Config.CHROMA_HOST,
    port=Config.CHROMA_PORT,
    collection_name=Config.CHROMA_COLLECTION
 )
 # Print configuration on startup
 if __name__ != '__main__':
    Config.print_config()
@@ -331,7 +340,11 @@ def crawl_rss_feed(feed_url, feed_name, feed_category='general', max_articles=10
        if not feed.entries:
            print(f"   ⚠ No entries found in feed")
-            return 0
+            return {
                'crawled': 0,
                'summarized': 0,
                'failed_summaries': 0
            }
        crawled_count = 0
        summarized_count = 0
@@ -388,6 +401,21 @@ def crawl_rss_feed(feed_url, feed_name, feed_category='general', max_articles=10
                        print(f"   ⚠ Summarization failed: {summary_result['error']}")
                        failed_summaries += 1
                # Extract keywords for personalization
                keywords_result = None
                if Config.OLLAMA_ENABLED and summary_result and summary_result['success']:
                    print(f"   🔑 Extracting keywords...")
                    keywords_result = ollama_client.extract_keywords(
                        original_title,
                        summary_result['summary'],
                        max_keywords=5
                    )
                    if keywords_result['success']:
                        print(f"   ✓ Keywords: {', '.join(keywords_result['keywords'])} ({keywords_result['duration']:.1f}s)")
                    else:
                        print(f"   ⚠ Keyword extraction failed: {keywords_result['error']}")
                # Prepare document
                article_doc = {
                    'title': original_title,
@@ -396,6 +424,7 @@ def crawl_rss_feed(feed_url, feed_name, feed_category='general', max_articles=10
                    'link': article_url,
                    'content': article_data.get('content', ''),  # Full article content
                    'summary': summary_result['summary'] if summary_result and summary_result['success'] else None,
                    'keywords': keywords_result['keywords'] if keywords_result and keywords_result['success'] else [],
                    'word_count': article_data.get('word_count', 0),
                    'summary_word_count': summary_result['summary_word_count'] if summary_result and summary_result['success'] else None,
                    'source': feed_name,
@@ -424,6 +453,17 @@ def crawl_rss_feed(feed_url, feed_name, feed_category='general', max_articles=10
                    crawled_count += 1
                    print(f"   ✓ Saved ({article_data.get('word_count', 0)} words)")
                    # Index in ChromaDB
                    try:
                        # Add mongo _id to article doc for reference
                        saved_article = articles_collection.find_one({'link': article_url})
                        if saved_article:
                            article_doc['_id'] = str(saved_article['_id'])
                            chroma_client.add_articles([article_doc])
                    except Exception as e:
                        print(f"   ⚠ Failed to index in ChromaDB: {e}")
                except DuplicateKeyError:
                    print(f"   ⚠ Duplicate key error")
                except Exception as e:
@@ -446,7 +486,11 @@ def crawl_rss_feed(feed_url, feed_name, feed_category='general', max_articles=10
    except Exception as e:
        print(f"   ✗ Error processing feed {feed_name}: {e}")
-        return 0
+        return {
            'crawled': 0,
            'summarized': 0,
            'failed_summaries': 0
        }
 def crawl_all_feeds(max_articles_per_feed=10):
--- a/news_crawler/ollama_client.py
+++ b/news_crawler/ollama_client.py
@@ -26,6 +26,43 @@ class OllamaClient:
        self.enabled = enabled
        self.timeout = timeout
    def _chat_request(self, messages, options=None):
        """
        Helper to make chat requests to Ollama
        Args:
            messages: List of message dicts [{'role': 'user', 'content': '...'}]
            options: Optional dict of model parameters
        Returns:
            str: Generated text content
        """
        if options is None:
            options = {}
        url = f"{self.base_url}/api/chat"
        headers = {'Content-Type': 'application/json'}
        if self.api_key:
            headers['Authorization'] = f'Bearer {self.api_key}'
        payload = {
            'model': self.model,
            'messages': messages,
            'stream': False,
            'options': options
        }
        response = requests.post(
            url,
            json=payload,
            headers=headers,
            timeout=self.timeout
        )
        response.raise_for_status()
        result = response.json()
        return result.get('message', {}).get('content', '').strip()
    def summarize_article(self, content, max_words=150):
        """
        Summarize article content using Ollama
@@ -70,37 +107,26 @@ class OllamaClient:
        start_time = time.time()
        try:
-            # Construct prompt
+            # Construct messages for chat API
-            prompt = self._build_summarization_prompt(content, max_words)
+            messages = [
-            
+                {
-            # Prepare request
+                    'role': 'system',
-            url = f"{self.base_url}/api/generate"
+                    'content': f"You are a skilled journalist writing for The New York Times. Summarize the provided article in English in {max_words} words or less.\\n\\nWrite in the clear, engaging, and authoritative style of New York Times Magazine:\\n- Lead with the most newsworthy information\\n- Use active voice and vivid language\\n- Make it accessible and easy to read\\n- Focus on what matters to readers\\n- Even if the source is in German or another language, write your summary entirely in English\\n\\nIMPORTANT: Write in plain text only. Do NOT use markdown formatting (no ##, **, *, bullets, etc.). Just write natural prose."
-            headers = {'Content-Type': 'application/json'}
+                },
-            if self.api_key:
+                {
-                headers['Authorization'] = f'Bearer {self.api_key}'
+                    'role': 'user',
-            
+                    'content': f"Summarize this article:\\n\\n{content}"
            payload = {
                'model': self.model,
                'prompt': prompt,
                'stream': False,
                'options': {
                    'temperature': 0.7,
                    'num_predict': 250  # Limit response length
                }
                }
            ]
-            # Make request
+            # Make request using chat endpoint
-            response = requests.post(
+            summary = self._chat_request(
-                url,
+                messages, 
-                json=payload,
+                options={
-                headers=headers,
+                    'temperature': 0.5,
-                timeout=self.timeout
+                    'num_predict': 350
                }
            )
            response.raise_for_status()
            # Parse response
            result = response.json()
            summary = result.get('response', '').strip()
            if not summary:
                return {
@@ -198,37 +224,26 @@ class OllamaClient:
        start_time = time.time()
        try:
-            # Construct prompt
+            # Construct messages for chat API
-            prompt = self._build_translation_prompt(title, target_language)
+            messages = [
-            
+                {
-            # Prepare request
+                    'role': 'system',
-            url = f"{self.base_url}/api/generate"
+                    'content': f"You are a professional translator. Translate the following German news headline to {target_language}.\\n\\nIMPORTANT: Provide ONLY the {target_language} translation. Do not include explanations, quotes, or any other text. Just the translated headline."
-            headers = {'Content-Type': 'application/json'}
+                },
-            if self.api_key:
+                {
-                headers['Authorization'] = f'Bearer {self.api_key}'
+                    'role': 'user',
-            
+                    'content': title
            payload = {
                'model': self.model,
                'prompt': prompt,
                'stream': False,
                'options': {
                    'temperature': 0.3,  # Lower temperature for consistent translations
                    'num_predict': 100   # Limit response length for title-length outputs
                }
                }
            ]
-            # Make request
+            # Make request using chat endpoint
-            response = requests.post(
+            translated_title = self._chat_request(
-                url,
+                messages,
-                json=payload,
+                options={
-                headers=headers,
+                    'temperature': 0.1,  # Low temperature for consistent translations
-                timeout=self.timeout
+                    'num_predict': 100   # Limit response length
                }
            )
            response.raise_for_status()
            # Parse response
            result = response.json()
            translated_title = result.get('response', '').strip()
            if not translated_title:
                return {
@@ -241,6 +256,13 @@ class OllamaClient:
            # Clean the translation output
            translated_title = self._clean_translation(translated_title)
            # Validate translation (if it's same as original, it might have failed)
            if translated_title.lower() == title.lower() and target_language == 'English':
                # Retry with more forceful prompt
                messages[0]['content'] += " If the text is already English, just output it as is."
                translated_title = self._chat_request(messages, options={'temperature': 0.1})
                translated_title = self._clean_translation(translated_title)
            return {
                'success': True,
                'translated_title': translated_title,
@@ -277,19 +299,6 @@ class OllamaClient:
                'duration': time.time() - start_time
            }
    def _build_translation_prompt(self, title, target_language):
        """Build prompt for title translation"""
        prompt = f"""Translate the following German news headline to {target_language}. 
 IMPORTANT: Provide ONLY the {target_language} translation. Do not include explanations, quotes, or any other text. Just the translated headline.
 German headline:
 {title}
 {target_language} translation:"""
        return prompt
    def _clean_translation(self, translation):
        """Clean translation output by removing quotes and extra text"""
        # Extract first line only
@@ -335,31 +344,6 @@ German headline:
        return text
    def _build_summarization_prompt(self, content, max_words):
        """Build prompt for article summarization"""
        # Truncate content if too long (keep first 5000 words)
        words = content.split()
        if len(words) > 5000:
            content = ' '.join(words[:5000]) + '...'
        prompt = f"""You are a skilled journalist writing for The New York Times. Summarize the following article in English in {max_words} words or less.
 Write in the clear, engaging, and authoritative style of New York Times Magazine:
 - Lead with the most newsworthy information
 - Use active voice and vivid language
 - Make it accessible and easy to read
 - Focus on what matters to readers
 - Even if the source is in German or another language, write your summary entirely in English
 IMPORTANT: Write in plain text only. Do NOT use markdown formatting (no ##, **, *, bullets, etc.). Just write natural prose.
 Article:
 {content}
 New York Times-style summary (max {max_words} words):"""
        return prompt
    def is_available(self):
        """
        Check if Ollama server is reachable
@@ -462,37 +446,24 @@ New York Times-style summary (max {max_words} words):"""
        start_time = time.time()
        try:
-            response = requests.post(
+            messages = [{'role': 'user', 'content': prompt}]
-                f"{self.base_url}/api/generate",
+            
-                json={
+            text = self._chat_request(
-                    "model": self.model,
+                messages,
-                    "prompt": prompt,
+                options={
                    "stream": False,
                    "options": {
                    "num_predict": max_tokens,
-                        "temperature": 0.1  # Low temperature for consistent answers
+                    "temperature": 0.1
                }
                },
                timeout=self.timeout
            )
            duration = time.time() - start_time
            if response.status_code == 200:
                result = response.json()
            return {
-                    'text': result.get('response', '').strip(),
+                'text': text,
                'success': True,
                'error': None,
                'duration': duration
            }
            else:
                return {
                    'text': '',
                    'success': False,
                    'error': f"HTTP {response.status_code}: {response.text}",
                    'duration': duration
                }
        except requests.exceptions.Timeout:
            return {
@@ -509,6 +480,89 @@ New York Times-style summary (max {max_words} words):"""
                'duration': time.time() - start_time
            }
    def extract_keywords(self, title, summary, max_keywords=5):
        """
        Extract keywords/topics from article for personalization
        Args:
            title: Article title
            summary: Article summary
            max_keywords: Maximum number of keywords to extract (default 5)
        Returns:
            {
                'keywords': list,      # List of extracted keywords
                'success': bool,       # Whether extraction succeeded
                'error': str or None,  # Error message if failed
                'duration': float      # Time taken in seconds
            }
        """
        if not self.enabled:
            return {
                'keywords': [],
                'success': False,
                'error': 'Ollama is disabled',
                'duration': 0
            }
        start_time = time.time()
        try:
            # Construct messages for chat API
            messages = [
                {
                    'role': 'system',
                    'content': f"Extract {max_keywords} key topics or keywords from the article.\\n\\nReturn ONLY the keywords separated by commas, nothing else. Focus on:\\n- Main topics (e.g., 'Bayern Munich', 'Oktoberfest', 'City Council')\\n- Locations (e.g., 'Marienplatz', 'Airport')\\n- Events or themes (e.g., 'Transportation', 'Housing', 'Technology')"
                },
                {
                    'role': 'user',
                    'content': f"Title: {title}\\nSummary: {summary}"
                }
            ]
            # Make request
            keywords_text = self._chat_request(
                messages,
                options={
                    'temperature': 0.2,
                    'num_predict': 100
                }
            )
            if not keywords_text:
                return {
                    'keywords': [],
                    'success': False,
                    'error': 'Ollama returned empty response',
                    'duration': time.time() - start_time
                }
            # Parse keywords from response
            keywords = [k.strip() for k in keywords_text.split(',')]
            keywords = [k for k in keywords if k and len(k) > 2][:max_keywords]
            return {
                'keywords': keywords,
                'success': True,
                'error': None,
                'duration': time.time() - start_time
            }
        except requests.exceptions.Timeout:
            return {
                'keywords': [],
                'success': False,
                'error': f"Request timed out after {self.timeout}s",
                'duration': time.time() - start_time
            }
        except Exception as e:
            return {
                'keywords': [],
                'success': False,
                'error': str(e),
                'duration': time.time() - start_time
            }
 if __name__ == '__main__':
    # Quick test
--- a/news_crawler/requirements.txt
+++ b/news_crawler/requirements.txt
@@ -7,3 +7,4 @@ python-dotenv==1.0.0
 schedule==1.2.0
 pytz==2023.3
 redis==5.0.1
 chromadb>=0.4.0
--- a/news_crawler/scheduled_crawler.py
+++ b/news_crawler/scheduled_crawler.py
@@ -37,12 +37,12 @@ def main():
    """Main scheduler loop"""
    print("🤖 Munich News Crawler Scheduler")
    print("="*60)
-    print("Schedule: Daily at 6:00 AM Berlin time")
+    print("Schedule: Every 3 hours")
    print("Timezone: Europe/Berlin (CET/CEST)")
    print("="*60)
-    # Schedule the crawler to run at 6 AM Berlin time
+    # Schedule the crawler to run every 3 hours
-    schedule.every().day.at("06:00").do(run_crawler)
+    schedule.every(3).hours.do(run_crawler)
    # Show next run time
    berlin_time = datetime.now(BERLIN_TZ)
--- a/news_sender/newsletter_template.html
+++ b/news_sender/newsletter_template.html
@@ -5,119 +5,113 @@
    <meta name="viewport" content="width=device-width, initial-scale=1.0">
    <meta http-equiv="X-UA-Compatible" content="IE=edge">
    <title>Munich News Daily</title>
    <!--[if mso]>
    <style type="text/css">
-        body, table, td {font-family: Arial, Helvetica, sans-serif !important;}
+        /* Client-specific resets */
-    </style>
+        body, table, td, a { -webkit-text-size-adjust: 100%; -ms-text-size-adjust: 100%; }
-    <![endif]-->
+        table, td { mso-table-lspace: 0pt; mso-table-rspace: 0pt; }
-</head>
+        img { -ms-interpolation-mode: bicubic; border: 0; height: auto; line-height: 100%; outline: none; text-decoration: none; }
 <body style="margin: 0; padding: 0; background-color: #f4f4f4; font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', Roboto, 'Helvetica Neue', Arial, sans-serif;">
    <!-- Wrapper Table -->
    <table role="presentation" width="100%" cellpadding="0" cellspacing="0" border="0" style="background-color: #f4f4f4;">
        <tr>
            <td align="center" style="padding: 20px 0;">
                <!-- Main Container -->
                <table role="presentation" width="600" cellpadding="0" cellspacing="0" border="0" style="background-color: #ffffff; max-width: 600px;">
-                    <!-- Header -->
+        /* General styles */
        body { height: 100% !important; margin: 0 !important; padding: 0 !important; width: 100% !important; }
        /* Hover effects */
        .hover-opacity:hover { opacity: 0.8 !important; }
        .read-more-btn:hover { background-color: #556cd6 !important; }
    </style>
 </head>
 <body style="margin: 0; padding: 0; background-color: #f0f2f5; font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', Roboto, Helvetica, Arial, sans-serif; color: #1a1a1a;">
    <table role="presentation" width="100%" cellpadding="0" cellspacing="0" border="0" style="background-color: #f0f2f5;">
        <tr>
-                        <td style="background-color: #1a1a1a; padding: 30px 40px; text-align: center;">
+            <td align="center" style="padding: 30px 10px;">
-                            <h1 style="margin: 0 0 8px 0; font-size: 28px; font-weight: 700; color: #ffffff; letter-spacing: -0.5px;">
+                
-                                Munich News Daily
+                <table role="presentation" width="600" cellpadding="0" cellspacing="0" border="0" style="max-width: 600px; width: 100%;">
-                            </h1>
+                    
-                            <p style="margin: 0; font-size: 14px; color: #999999; letter-spacing: 0.5px;">
+                    <tr>
                        <td align="center" style="padding-bottom: 25px;">
                            <p style="margin: 0 0 5px 0; font-size: 12px; font-weight: 700; letter-spacing: 1.5px; text-transform: uppercase; color: #667eea;">
                                {{ date }}
                            </p>
                            <h1 style="margin: 0; font-size: 32px; font-weight: 800; letter-spacing: -1px; color: #1a1a1a;">
                                Munich News Daily
                            </h1>
                        </td>
                    </tr>
                    <!-- Greeting -->
                    <tr>
-                        <td style="padding: 30px 40px 20px 40px;">
+                        <td style="background-color: #ffffff; border-radius: 16px; padding: 30px; margin-bottom: 20px; box-shadow: 0 2px 4px rgba(0,0,0,0.05);">
-                            <p style="margin: 0; font-size: 16px; line-height: 1.5; color: #333333;">
+                            <p style="margin: 0 0 10px 0; font-size: 18px; font-weight: 600; color: #1a1a1a;">
                                Good morning ☀️
                            </p>
-                            <p style="margin: 15px 0 0 0; font-size: 15px; line-height: 1.6; color: #666666; text-align: justify;">
+                            <p style="margin: 0 0 25px 0; font-size: 16px; line-height: 1.6; color: #555555;">
-                                Here's what's happening in Munich today. We've summarized {{ article_count }} stories using AI so you can stay informed in under 5 minutes.
+                                Here is your AI-curated briefing. We've summarized <strong>{{ article_count }} stories</strong> to get you up to speed in under 5 minutes.
                            </p>
                            {% if weather and weather.success %}
-                            <!-- Weather Widget -->
+                            <table role="presentation" width="100%" cellpadding="0" cellspacing="0" border="0" style="background: linear-gradient(135deg, #667eea 0%, #764ba2 100%); border-radius: 12px; overflow: hidden;">
                            <table role="presentation" width="100%" cellpadding="0" cellspacing="0" border="0" style="margin-top: 20px; background: linear-gradient(135deg, #667eea 0%, #764ba2 100%); border-radius: 12px; overflow: hidden;">
                                <tr>
                                    <td style="padding: 20px;">
                                        <table role="presentation" width="100%" cellpadding="0" cellspacing="0" border="0">
                                            <tr>
                                                <td style="width: 60%; vertical-align: middle;">
                                                    <p style="margin: 0 0 5px 0; font-size: 13px; color: rgba(255,255,255,0.9); font-weight: 600;">
                                                        TODAY'S WEATHER
                                                    </p>
                                                    <p style="margin: 0; font-size: 32px; color: #ffffff; font-weight: 700; line-height: 1;">
                                                        {{ weather.icon }} {{ weather.temperature }}°C
                                                    </p>
                                                    <p style="margin: 5px 0 0 0; font-size: 14px; color: rgba(255,255,255,0.9);">
                                                        {{ weather.condition }}
                                                    </p>
                                                </td>
                                                <td style="width: 40%; text-align: right; vertical-align: middle;">
                                                    <p style="margin: 0; font-size: 14px; color: rgba(255,255,255,0.9);">
                                                        High: <strong style="color: #ffffff;">{{ weather.high }}°C</strong>
                                                    </p>
                                                    <p style="margin: 5px 0 0 0; font-size: 14px; color: rgba(255,255,255,0.9);">
                                                        Low: <strong style="color: #ffffff;">{{ weather.low }}°C</strong>
                                                    </p>
                                                </td>
                                            </tr>
                                        </table>
                                    </td>
                                </tr>
                            </table>
                            {% endif %}
                        </td>
                    </tr>
                    <!-- Divider -->
                    <tr>
                        <td style="padding: 0 40px;">
                            <div style="height: 1px; background-color: #e0e0e0;"></div>
                        </td>
                    </tr>
                    <!-- TL;DR Section -->
                    <tr>
                        <td style="padding: 30px 40px;">
                            <h2 style="margin: 0 0 8px 0; font-size: 22px; font-weight: 700; color: #1a1a1a;">
                                📋 TL;DR - Quick Summary
                            </h2>
                            <p style="margin: 0 0 20px 0; font-size: 13px; color: #999999;">
                                Scan through today's top stories in seconds
                            </p>
                            <!-- TL;DR Box -->
                            <table role="presentation" width="100%" cellpadding="0" cellspacing="0" border="0" style="background-color: #f8f9fa; border-radius: 8px; border-left: 4px solid #667eea;">
                                <tr>
                                    <td style="padding: 20px 25px;">
                                        <table role="presentation" width="100%" cellpadding="0" cellspacing="0" border="0">
                                            <tr>
                                                <td style="vertical-align: middle;">
                                                    <div style="font-size: 36px; font-weight: 700; color: #ffffff; line-height: 1;">
                                                        {{ weather.icon }} {{ weather.temperature }}°
                                                    </div>
                                                    <div style="font-size: 14px; color: rgba(255,255,255,0.9); margin-top: 5px; font-weight: 500;">
                                                        {{ weather.condition }}
                                                    </div>
                                                </td>
                                                <td align="right" style="vertical-align: middle;">
                                                    <table role="presentation" cellpadding="0" cellspacing="0" border="0">
                                                        <tr>
                                                            <td style="padding-right: 15px; border-right: 1px solid rgba(255,255,255,0.3);">
                                                                <div style="font-size: 12px; color: rgba(255,255,255,0.8); text-transform: uppercase;">High</div>
                                                                <div style="font-size: 16px; font-weight: 700; color: #ffffff;">{{ weather.high }}°</div>
                                                            </td>
                                                            <td style="padding-left: 15px;">
                                                                <div style="font-size: 12px; color: rgba(255,255,255,0.8); text-transform: uppercase;">Low</div>
                                                                <div style="font-size: 16px; font-weight: 700; color: #ffffff;">{{ weather.low }}°</div>
                                                            </td>
                                                        </tr>
                                                    </table>
                                                </td>
                                            </tr>
                                        </table>
                                    </td>
                                </tr>
                            </table>
                            {% endif %}
                        </td>
                    </tr>
                    <tr><td style="height: 20px;"></td></tr>
                    <tr>
                        <td style="background-color: #ffffff; border-radius: 16px; padding: 30px; box-shadow: 0 2px 4px rgba(0,0,0,0.05);">
                            <table role="presentation" width="100%" cellpadding="0" cellspacing="0" border="0">
                                <tr>
                                    <td style="padding-bottom: 20px; border-bottom: 2px solid #f0f2f5;">
                                        <h2 style="margin: 0; font-size: 20px; font-weight: 700; color: #1a1a1a;">
                                            ⚡️ Quick Summary
                                        </h2>
                                    </td>
                                </tr>
                                <tr>
                                    <td style="padding-top: 20px;">
                                        {% for section in category_sections %}
                                        {% for article in section.articles %}
-                                        <table role="presentation" width="100%" cellpadding="0" cellspacing="0" border="0" style="margin-bottom: 15px;">
+                                        <table role="presentation" width="100%" cellpadding="0" cellspacing="0" border="0" style="margin-bottom: 12px;">
                                            <tr>
-                                                <td style="width: 30px; vertical-align: top; padding-top: 2px;">
+                                                <td style="width: 24px; vertical-align: top; padding-top: 4px;">
-                                                    <span style="display: inline-block; background-color: #667eea; color: #ffffff; width: 22px; height: 22px; line-height: 22px; text-align: center; border-radius: 50%; font-size: 11px; font-weight: 700;">
+                                                    <div style="height: 6px; width: 6px; background-color: #667eea; border-radius: 50%;"></div>
                                                        {{ loop.index }}
                                                    </span>
                                                </td>
                                                <td style="vertical-align: top;">
-                                                    <p style="margin: 0; font-size: 14px; line-height: 1.6; color: #333333;">
+                                                    <p style="margin: 0; font-size: 15px; line-height: 1.5; color: #4a4a4a;">
-                                                        <strong style="color: #1a1a1a;">{{ article.title_en if article.title_en else article.title }}</strong>
+                                                        <strong style="color: #1a1a1a;">{{ article.title_en if article.title_en else article.title }}</strong> — {{ article.summary.split('.')[0] }}.
                                                        <br>
                                                        <span style="color: #666666;">{{ article.summary.split('.')[0] }}.</span>
                                                    </p>
                                                </td>
                                            </tr>
                                        </table>
                                        {% if not loop.last %}
                                        <div style="height: 1px; background-color: #e0e0e0; margin: 10px 0;"></div>
                                        {% endif %}
                                        {% endfor %}
                                        {% endfor %}
                                    </td>
@@ -127,88 +121,48 @@
                    </tr>
                    {% if transport_disruptions and transport_disruptions|length > 0 %}
-                    <!-- Divider -->
+                    <tr><td style="height: 20px;"></td></tr>
                    <tr>
-                        <td style="padding: 0 40px;">
+                        <td style="background-color: #ffffff; border-radius: 16px; padding: 30px; box-shadow: 0 2px 4px rgba(0,0,0,0.05);">
-                            <div style="height: 2px; background-color: #e0e0e0;"></div>
+                            <h2 style="margin: 0 0 15px 0; font-size: 20px; font-weight: 700; color: #1a1a1a;">
-                        </td>
+                                🚆 S-Bahn Updates
                    </tr>
                    <!-- Transport Disruptions Section -->
                    <tr>
                        <td style="padding: 30px 40px;">
                            <h2 style="margin: 0 0 20px 0; font-size: 22px; font-weight: 700; color: #1a1a1a;">
                                🚆 S-Bahn Disruptions Today
                            </h2>
                            <p style="margin: 0 0 20px 0; font-size: 14px; color: #666666;">
                                Current service disruptions affecting Munich S-Bahn:
                            </p>
                            {% for disruption in transport_disruptions %}
-                            <!-- Disruption Card -->
+                            <table role="presentation" width="100%" cellpadding="0" cellspacing="0" border="0" style="background-color: #fff8f0; border-radius: 8px; border: 1px solid #ffeeba; margin-bottom: 10px;">
                            <table role="presentation" width="100%" cellpadding="0" cellspacing="0" border="0" style="margin-bottom: 15px; background-color: #fff8f0; border-left: 4px solid #ff9800; border-radius: 4px;">
                                <tr>
-                                    <td style="padding: 15px 20px;">
+                                    <td style="padding: 15px;">
-                                        <!-- Severity and Lines -->
+                                        <p style="margin: 0 0 5px 0; font-size: 12px; font-weight: 700; color: #c05621; text-transform: uppercase;">
-                                        <p style="margin: 0 0 8px 0; font-size: 13px; color: #666666;">
+                                            {{ disruption.severity_icon }} {{ disruption.lines_str }}
                                            {{ disruption.severity_icon }} <strong style="color: #000000;">{{ disruption.lines_str }}</strong>
                                        </p>
-                                        
+                                        <p style="margin: 0 0 5px 0; font-size: 15px; font-weight: 700; color: #2d3748;">
                                        <!-- Title -->
                                        <p style="margin: 0 0 8px 0; font-size: 15px; font-weight: 700; color: #1a1a1a; line-height: 1.4;">
                                            {{ disruption.title }}
                                        </p>
                                        <!-- Description -->
                                        {% if disruption.description %}
-                                        <p style="margin: 0 0 8px 0; font-size: 14px; color: #333333; line-height: 1.5;">
+                                        <p style="margin: 0; font-size: 14px; color: #4a5568; line-height: 1.4;">
                                            {{ disruption.description }}
                                        </p>
                                        {% endif %}
                                        <!-- Time -->
                                        {% if disruption.start_time_str or disruption.end_time_str %}
                                        <p style="margin: 0; font-size: 13px; color: #666666;">
                                            ⏰ 
                                            {% if disruption.start_time_str %}
                                            From {{ disruption.start_time_str }}
                                            {% endif %}
                                            {% if disruption.end_time_str %}
                                            until {{ disruption.end_time_str }}
                                            {% endif %}
                                        </p>
                                        {% endif %}
                                    </td>
                                </tr>
                            </table>
                            {% endfor %}
-                            
+                            <p style="margin: 10px 0 0 0; font-size: 13px; color: #718096;">
-                            <p style="margin: 15px 0 0 0; font-size: 12px; color: #999999; font-style: italic;">
+                                Check <a href="https://www.mvg.de" style="color: #667eea; text-decoration: underline;">MVG.de</a> for live times.
                                💡 Plan your commute accordingly. Check <a href="https://www.mvg.de" style="color: #667eea; text-decoration: none;">MVG.de</a> for real-time updates.
                            </p>
                        </td>
                    </tr>
                    {% endif %}
                    <!-- Divider -->
                    <tr>
                        <td style="padding: 0 40px;">
                            <div style="height: 2px; background-color: #e0e0e0;"></div>
                        </td>
                    </tr>
                    <!-- Category Sections -->
                    {% for section in category_sections %}
                    <tr>
-                        <td style="padding: 30px 40px 15px 40px;">
+                        <td align="center" style="padding: 40px 0 20px 0;">
-                            <table role="presentation" width="100%" cellpadding="0" cellspacing="0" border="0">
+                            <table role="presentation" border="0" cellspacing="0" cellpadding="0">
                                <tr>
-                                    <td>
+                                    <td align="center" style="background-color: #e2e8f0; color: #4a5568; border-radius: 20px; padding: 8px 20px;">
-                                        <h2 style="margin: 0; font-size: 22px; font-weight: 700; color: #1a1a1a;">
+                                        <p style="margin: 0; font-size: 13px; font-weight: 700; letter-spacing: 0.5px; text-transform: uppercase;">
-                                            {{ section.icon }} {{ section.name }}
+                                            {{ section.icon }} &nbsp; {{ section.name }}
                                        </h2>
                                        <p style="margin: 8px 0 0 0; font-size: 13px; color: #666666;">
                                            Top stories in {{ section.name.lower() }}
                                        </p>
                                    </td>
                                </tr>
@@ -216,155 +170,99 @@
                        </td>
                    </tr>
                    <!-- Category Articles -->
                    {% for article in section.articles %}
                    <tr>
-                        <td style="padding: 25px 40px;">
+                        <td style="background-color: #ffffff; border-radius: 16px; padding: 30px; margin-bottom: 20px; box-shadow: 0 2px 4px rgba(0,0,0,0.05);">
-                            <!-- Article Number Badge -->
+                            <p style="margin: 0 0 12px 0; font-size: 12px; color: #718096; font-weight: 600; text-transform: uppercase; letter-spacing: 0.5px;">
                            <table role="presentation" width="100%" cellpadding="0" cellspacing="0" border="0">
                                <tr>
                                    <td>
                                        <span style="display: inline-block; background-color: #000000; color: #ffffff; width: 24px; height: 24px; line-height: 24px; text-align: center; border-radius: 50%; font-size: 12px; font-weight: 600;">
                                            {{ loop.index }}
                                        </span>
                                    </td>
                                </tr>
                            </table>
                            <!-- Article Title -->
                            <h2 style="margin: 12px 0 8px 0; font-size: 19px; font-weight: 700; line-height: 1.3; color: #1a1a1a;">
                                {{ article.title_en if article.title_en else article.title }}
                            </h2>
                            <!-- Original German Title (subtitle) -->
                            {% if article.title_en and article.title_en != article.title %}
                            <p style="margin: 0 0 12px 0; font-size: 13px; color: #999999; font-style: italic;">
                                Original: {{ article.title }}
                            </p>
                            {% endif %}
                            <!-- Article Meta -->
                            <p style="margin: 0 0 12px 0; font-size: 13px; color: #999999;">
                                {% if article.is_clustered %}
-                                <span style="color: #000000; font-weight: 600;">Multiple sources</span>
+                                    Multiple Sources
                                {% else %}
-                                <span style="color: #000000; font-weight: 600;">{{ article.source }}</span>
+                                    {{ article.source }}
                                {% if article.author %}
                                <span> • {{ article.author }}</span>
                                {% endif %}
                                {% endif %}
                            </p>
-                            <!-- Article Summary -->
+                            <h3 style="margin: 0 0 10px 0; font-size: 22px; font-weight: 700; line-height: 1.3; color: #1a1a1a;">
-                            <p style="margin: 0 0 15px 0; font-size: 15px; line-height: 1.6; color: #333333; text-align: justify;">
+                                <a href="{{ article.link }}" style="color: #1a1a1a; text-decoration: none;">
                                    {{ article.title_en if article.title_en else article.title }}
                                </a>
                            </h3>
                            {% if article.title_en and article.title_en != article.title %}
                            <p style="margin: 0 0 15px 0; font-size: 13px; color: #a0aec0; font-style: italic;">
                                "{{ article.title }}"
                            </p>
                            {% endif %}
                            <p style="margin: 0 0 20px 0; font-size: 16px; line-height: 1.6; color: #4a5568;">
                                {{ article.summary }}
                            </p>
-                            <!-- Read More Links -->
+                            <table role="presentation" width="100%" cellpadding="0" cellspacing="0" border="0">
                                <tr>
                                    <td>
                                        {% if article.is_clustered and article.sources %}
-                            <!-- Multiple sources -->
+                                            <p style="margin: 0; font-size: 13px; color: #718096;">
-                            <p style="margin: 0 0 8px 0; font-size: 13px; color: #666666;">
+                                                <strong>Read full coverage:</strong><br>
                                📰 Covered by {{ article.article_count }} sources:
                            </p>
                            <div style="margin: 0;">
                                                {% for source in article.sources %}
-                                <a href="{{ source.link }}" style="display: inline-block; color: #000000; text-decoration: none; font-size: 13px; font-weight: 600; border-bottom: 2px solid #000000; padding-bottom: 2px; margin-right: 15px; margin-bottom: 8px;">
+                                                <a href="{{ source.link }}" style="color: #667eea; text-decoration: none; margin-right: 10px; display: inline-block; margin-top: 5px;">
-                                    {{ source.name }} →
+                                                    {{ source.name }} ↗
                                                </a>
                                                {% endfor %}
                            </div>
                            {% else %}
                            <!-- Single source -->
                            <a href="{{ article.link }}" style="display: inline-block; color: #000000; text-decoration: none; font-size: 14px; font-weight: 600; border-bottom: 2px solid #000000; padding-bottom: 2px;">
                                Read more →
                            </a>
                            {% endif %}
                        </td>
                    </tr>
                    <!-- Article Divider -->
                    {% if not loop.last %}
                    <tr>
                        <td style="padding: 0 40px;">
                            <div style="height: 1px; background-color: #f0f0f0;"></div>
                        </td>
                    </tr>
                    {% endif %}
                    {% endfor %}
                    <!-- Category Section Divider -->
                    {% if not loop.last %}
                    <tr>
                        <td style="padding: 25px 40px;">
                            <div style="height: 2px; background-color: #e0e0e0;"></div>
                        </td>
                    </tr>
                    {% endif %}
                    {% endfor %}
                    <!-- Bottom Divider -->
                    <tr>
                        <td style="padding: 25px 40px 0 40px;">
                            <div style="height: 1px; background-color: #e0e0e0;"></div>
                        </td>
                    </tr>
                        <td style="padding: 30px 40px;">
                            <table role="presentation" width="100%" cellpadding="0" cellspacing="0" border="0" style="background-color: #f8f8f8; border-radius: 8px;">
                                <tr>
                                    <td style="padding: 25px; text-align: center;">
                                        <p style="margin: 0 0 8px 0; font-size: 13px; color: #666666; text-transform: uppercase; letter-spacing: 1px; font-weight: 600;">
                                            Today's Digest
                                            </p>
-                                        <p style="margin: 0; font-size: 36px; font-weight: 700; color: #000000;">
+                                        {% else %}
                                            <table role="presentation" cellpadding="0" cellspacing="0" border="0">
                                                <tr>
                                                    <td style="background-color: #f0f2f5; border-radius: 8px; padding: 10px 20px;">
                                                        <a href="{{ article.link }}" style="color: #1a1a1a; font-size: 14px; font-weight: 600; text-decoration: none; display: block;">
                                                            Read full story &rarr;
                                                        </a>
                                                    </td>
                                                </tr>
                                            </table>
                                        {% endif %}
                                    </td>
                                </tr>
                            </table>
                        </td>
                    </tr>
                    <tr><td style="height: 20px;"></td></tr>
                    {% endfor %}
                    {% endfor %}
                    <tr>
                        <td align="center" style="padding: 40px 0;">
                            <p style="margin: 0 0 5px 0; font-size: 48px; font-weight: 800; color: #cbd5e0; line-height: 1;">
                                {{ article_count }}
                            </p>
-                                        <p style="margin: 8px 0 0 0; font-size: 14px; color: #666666;">
+                            <p style="margin: 0; font-size: 14px; font-weight: 600; color: #a0aec0; text-transform: uppercase; letter-spacing: 1px;">
-                                            stories • AI-summarized • 5 min read
+                                Stories Summarized
                            </p>
                        </td>
                    </tr>
-                            </table>
+
                        </td>
                    <!-- Footer -->
                    <tr>
-                        <td style="background-color: #1a1a1a; padding: 30px 40px; text-align: center;">
+                        <td style="text-align: center; padding: 0 20px 40px 20px;">
-                            <p style="margin: 0 0 15px 0; font-size: 14px; color: #ffffff; font-weight: 600;">
+                            <p style="margin: 0 0 20px 0; font-size: 13px; line-height: 1.5; color: #718096;">
-                                Munich News Daily
+                                <strong>Munich News Daily</strong><br>
-                            </p>
+                                AI-powered news for busy locals.
                            <p style="margin: 0 0 20px 0; font-size: 13px; color: #999999; line-height: 1.5;">
                                AI-powered news summaries for busy people.<br>
                                Delivered daily to your inbox.
                            </p>
-                            <!-- Footer Links -->
+                            <p style="margin: 0 0 20px 0; font-size: 12px; color: #a0aec0;">
-                            <p style="margin: 0; font-size: 12px; color: #666666;">
+                                <a href="{{ website_link }}" style="color: #718096; text-decoration: underline;">Website</a> &nbsp;•&nbsp; 
-                                <a href="{{ website_link }}" style="color: #999999; text-decoration: none;">Visit Website</a>
+                                <a href="{{ preferences_link }}" style="color: #718096; text-decoration: underline;">Preferences</a> &nbsp;•&nbsp; 
-                                <span style="color: #444444;"> • </span>
+                                <a href="{{ unsubscribe_link }}" style="color: #718096; text-decoration: underline;">Unsubscribe</a>
                                <a href="{{ preferences_link }}" style="color: #999999; text-decoration: none;">Manage Preferences</a>
                                <span style="color: #444444;"> • </span>
                                <a href="{{ unsubscribe_link }}" style="color: #999999; text-decoration: none;">Unsubscribe</a>
                            </p>
-                            {% if tracking_enabled %}
+                            <p style="margin: 0; font-size: 11px; color: #cbd5e0;">
-                            <!-- Privacy Notice -->
+                                © {{ year }} Munich News Daily. Made with 🥨 in Bavaria.
                            <p style="margin: 20px 0 0 0; font-size: 11px; color: #666666; line-height: 1.4;">
                                This email contains tracking to measure engagement and improve our content.<br>
                                We respect your privacy and anonymize data after 90 days.
                            </p>
                            {% endif %}
                            <p style="margin: 20px 0 0 0; font-size: 11px; color: #666666;">
                                © {{ year }} Munich News Daily. All rights reserved.
                            </p>
                        </td>
                    </tr>
                </table>
                <!-- End Main Container -->
            </td>
        </tr>
    </table>
-    <!-- End Wrapper Table -->
+
 </body>
 </html>
--- a/news_sender/sender_service.py
+++ b/news_sender/sender_service.py
@@ -166,8 +166,12 @@ def get_latest_articles_by_categories(categories=None, articles_per_category=3,
    # Get cluster summaries collection
    cluster_summaries_collection = db['cluster_summaries']
-    # If no categories specified, get all available categories
+    # If no categories specified, get all available categories from database
    if categories is None:
        # Dynamically get categories from articles collection
        categories = list(articles_collection.distinct('category'))
        if not categories:
            # Fallback to default categories if no articles exist yet
            categories = ['general', 'local', 'sports', 'science']
    articles = []
@@ -176,6 +180,7 @@ def get_latest_articles_by_categories(categories=None, articles_per_category=3,
    # Fetch articles for each category separately
    for category in categories:
        # Query for articles in this category from today
        # Fetch more than needed to allow for source diversity filtering
        cursor = articles_collection.find({
            'summary': {'$exists': True, '$ne': None},
            'category': category,
@@ -183,9 +188,10 @@ def get_latest_articles_by_categories(categories=None, articles_per_category=3,
                {'published_at': {'$gte': today_start}},
                {'created_at': {'$gte': today_start}}
            ]
-        }).sort('created_at', -1).limit(articles_per_category)
+        }).sort('created_at', -1).limit(articles_per_category * 3)  # Fetch 3x to allow diversity
        category_articles = []
        source_count = {}  # Track how many articles from each source
        for doc in cursor:
            # Double-check the date to ensure it's from today
@@ -268,15 +274,71 @@ def get_latest_articles_by_categories(categories=None, articles_per_category=3,
                    'is_clustered': False
                })
        # Diversify sources: prioritize articles from different sources
        # Sort by: clustered first, then by source diversity (fewer articles from same source)
        diversified_articles = []
        source_usage = {}
        # First pass: add clustered articles (they represent multiple sources)
        for article in category_articles:
            if article.get('is_clustered'):
                diversified_articles.append(article)
        # Second pass: add non-clustered articles with source diversity
        for article in category_articles:
            if not article.get('is_clustered'):
                source = article.get('source', 'unknown')
                # Prefer sources we haven't used much yet
                if source not in source_usage:
                    source_usage[source] = 0
                # Add article and track source usage
                diversified_articles.append(article)
                source_usage[source] += 1
        # Sort by source diversity: clustered first, then by how many times we've used this source
        diversified_articles.sort(key=lambda x: (
            0 if x.get('is_clustered') else 1,  # Clustered first
            -x.get('article_count', 1),  # More sources in cluster = higher priority
            source_usage.get(x.get('source', 'unknown'), 0)  # Fewer from same source = higher priority
        ))
        # Take only the requested number per category
        category_articles = diversified_articles[:articles_per_category]
        # Add this category's articles to the main list
        articles.extend(category_articles)
-    # Sort articles: clustered articles first (by source count), then by recency
+    # Final sort with source diversity across all categories
-    # This prioritizes stories covered by multiple sources
+    # Prioritize: 1) Clustered articles, 2) Source diversity, 3) Recency
-    articles.sort(key=lambda x: (
+    import random
-        -1 if x.get('is_clustered') else 0,  # Clustered first
+    
-        -x.get('article_count', 1),  # More sources = higher priority
+    # Group by clustered vs non-clustered
-    ), reverse=True)
+    clustered = [a for a in articles if a.get('is_clustered')]
    non_clustered = [a for a in articles if not a.get('is_clustered')]
    # Sort clustered by article count (more sources = more important)
    clustered.sort(key=lambda x: -x.get('article_count', 1))
    # For non-clustered, shuffle within each category to add variety
    # This prevents the same sources from always appearing first
    from collections import defaultdict
    by_category = defaultdict(list)
    for article in non_clustered:
        by_category[article.get('category', 'general')].append(article)
    # Shuffle each category's articles to mix sources
    for cat_articles in by_category.values():
        random.shuffle(cat_articles)
    # Reconstruct non-clustered list with shuffled articles
    non_clustered = []
    # Use all categories that exist in the articles
    for cat in sorted(by_category.keys()):
        non_clustered.extend(by_category[cat])
    # Combine: clustered first, then shuffled non-clustered
    articles = clustered + non_clustered
    return articles
@@ -345,7 +407,8 @@ def render_newsletter_html(articles, subscriber_categories=None, tracking_enable
        'general': {'name': 'Top Trending', 'icon': '🔥'},
        'local': {'name': 'Local Events', 'icon': '🏛️'},
        'sports': {'name': 'Sports', 'icon': '⚽'},
-        'science': {'name': 'Science & Tech', 'icon': '🔬'}
+        'science': {'name': 'Science & Tech', 'icon': '🔬'},
        'business': {'name': 'Business', 'icon': '💼'}
    }
    for category, category_articles in sorted(articles_by_category.items()):
--- a/tests/backend/test_personalization_system.py
+++ b/tests/backend/test_personalization_system.py
@@ -0,0 +1,221 @@
 #!/usr/bin/env python3
 """
 Comprehensive test suite for the personalization system.
 Tests all 4 phases: keyword extraction, click tracking, interest profiling, and personalization.
 """
 import sys
 from pymongo import MongoClient
 from datetime import datetime
 # Import services
 from services.tracking_service import create_newsletter_tracking
 from services.interest_profiling_service import (
    update_user_interests,
    get_user_interests,
    get_top_interests,
    build_interests_from_history
 )
 from services.personalization_service import (
    calculate_article_score,
    rank_articles_for_user,
    select_personalized_articles,
    get_personalization_stats
 )
 from config import Config
 # Connect to MongoDB
 client = MongoClient(Config.MONGODB_URI)
 db = client[Config.DB_NAME]
 articles_collection = db['articles']
 link_clicks_collection = db['link_clicks']
 user_interests_collection = db['user_interests']
 def test_phase1_keywords():
    """Phase 1: Verify articles have keywords extracted"""
    print("\n" + "="*60)
    print("Phase 1: Keyword Extraction")
    print("="*60)
    articles_with_keywords = articles_collection.count_documents({
        'keywords': {'$exists': True, '$ne': []}
    })
    if articles_with_keywords == 0:
        print("❌ No articles with keywords found")
        print("   Run a crawl first to extract keywords")
        return False
    sample = articles_collection.find_one({'keywords': {'$exists': True, '$ne': []}})
    print(f"✓ Found {articles_with_keywords} articles with keywords")
    print(f"  Sample: {sample.get('title', 'N/A')[:50]}...")
    print(f"  Keywords: {sample.get('keywords', [])[:3]}")
    return True
 def test_phase2_tracking():
    """Phase 2: Verify tracking includes keywords and metadata"""
    print("\n" + "="*60)
    print("Phase 2: Click Tracking Enhancement")
    print("="*60)
    test_email = 'test-phase2@example.com'
    # Clean up
    link_clicks_collection.delete_many({'subscriber_email': test_email})
    # Get article with keywords
    article = articles_collection.find_one({'keywords': {'$exists': True, '$ne': []}})
    if not article:
        print("❌ No articles found")
        return False
    # Create tracking
    tracking_data = create_newsletter_tracking(
        newsletter_id='test-phase2',
        subscriber_email=test_email,
        article_links=[{
            'url': article['link'],
            'title': article.get('title', '')
        }]
    )
    # Verify tracking record
    tracking_id = list(tracking_data['link_tracking_map'].values())[0]
    tracking_record = link_clicks_collection.find_one({'tracking_id': tracking_id})
    has_metadata = (
        tracking_record.get('article_id') is not None and
        tracking_record.get('category') is not None and
        len(tracking_record.get('keywords', [])) > 0
    )
    # Clean up
    link_clicks_collection.delete_many({'subscriber_email': test_email})
    db['newsletter_sends'].delete_many({'subscriber_email': test_email})
    if has_metadata:
        print(f"✓ Tracking records include metadata")
        print(f"  Article ID: {tracking_record.get('article_id')}")
        print(f"  Category: {tracking_record.get('category')}")
        print(f"  Keywords: {len(tracking_record.get('keywords', []))} keywords")
        return True
    else:
        print("❌ Tracking records missing metadata")
        return False
 def test_phase3_profiling():
    """Phase 3: Verify interest profiles are built from clicks"""
    print("\n" + "="*60)
    print("Phase 3: User Interest Profiling")
    print("="*60)
    test_email = 'test-phase3@example.com'
    # Clean up
    user_interests_collection.delete_many({'email': test_email})
    # Create profile
    update_user_interests(test_email, ['Bayern Munich', 'Football'], 'sports')
    update_user_interests(test_email, ['Transportation', 'Munich'], 'local')
    # Verify profile
    profile = get_user_interests(test_email)
    # Clean up
    user_interests_collection.delete_many({'email': test_email})
    if profile and profile['total_clicks'] == 2:
        print(f"✓ Interest profile created")
        print(f"  Total clicks: {profile['total_clicks']}")
        print(f"  Categories: {len(profile.get('categories', {}))}")
        print(f"  Keywords: {len(profile.get('keywords', {}))}")
        return True
    else:
        print("❌ Interest profile not created correctly")
        return False
 def test_phase4_personalization():
    """Phase 4: Verify articles are ranked by user interests"""
    print("\n" + "="*60)
    print("Phase 4: Personalized Newsletter Generation")
    print("="*60)
    test_email = 'test-phase4@example.com'
    # Clean up
    user_interests_collection.delete_many({'email': test_email})
    # Get articles
    articles = list(articles_collection.find(
        {'keywords': {'$exists': True, '$ne': []}},
        limit=5
    ))
    if len(articles) < 3:
        print("❌ Not enough articles found")
        return False
    # Create profile
    update_user_interests(test_email, ['Bayern Munich', 'Football'], 'sports')
    # Rank articles
    ranked = rank_articles_for_user(articles, test_email)
    # Select personalized
    selected = select_personalized_articles(articles, test_email, max_articles=3)
    # Clean up
    user_interests_collection.delete_many({'email': test_email})
    has_scores = all('personalization_score' in a for a in selected)
    if has_scores and len(selected) > 0:
        print(f"✓ Articles ranked and selected")
        print(f"  Total ranked: {len(ranked)}")
        print(f"  Selected: {len(selected)}")
        print(f"  Top score: {selected[0].get('personalization_score', 0):.3f}")
        return True
    else:
        print("❌ Personalization failed")
        return False
 def main():
    """Run all personalization tests"""
    print("\n" + "="*60)
    print("PERSONALIZATION SYSTEM TEST SUITE")
    print("="*60)
    results = {
        'Phase 1: Keyword Extraction': test_phase1_keywords(),
        'Phase 2: Click Tracking': test_phase2_tracking(),
        'Phase 3: Interest Profiling': test_phase3_profiling(),
        'Phase 4: Personalization': test_phase4_personalization()
    }
    print("\n" + "="*60)
    print("TEST RESULTS")
    print("="*60)
    for phase, passed in results.items():
        status = "✅ PASS" if passed else "❌ FAIL"
        print(f"{status} - {phase}")
    all_passed = all(results.values())
    if all_passed:
        print("\n🎉 All personalization tests PASSED!")
        return 0
    else:
        print("\n❌ Some tests FAILED")
        return 1
 if __name__ == '__main__':
    sys.exit(main())
Author	SHA1	Message	Date
dongho	7346ee9de2	update All checks were successful dongho-repo/Munich-news/pipeline/head This commit looks good Details	2025-12-10 15:57:07 +00:00
dongho	6e9fbe44c4	update	2025-12-10 15:52:41 +00:00
dongho	4e8b60f77c	update	2025-12-10 15:50:11 +00:00
dongho	50b9888004	update chromadb version	2025-12-10 12:46:47 +00:00
dongho	6c8d6d0940	Add ChromaDB implementation	2025-12-10 12:46:17 +00:00
dongho	57f37c8dc0	update	2025-12-10 12:43:18 +00:00
dongho	19fabacf5a	update	2025-11-28 15:50:44 +00:00
Dongho Kim	1656f19077	update	2025-11-21 12:29:41 +01:00
Dongho Kim	1c0926a7f0	update	2025-11-20 19:15:41 +01:00
Dongho Kim	ab47e518fa	update	2025-11-20 12:40:56 +01:00
Dongho Kim	e52042d5a8	update	2025-11-20 12:38:29 +01:00
Dongho Kim	2dc6bf44f8	update	2025-11-20 12:32:48 +01:00
Dongho Kim	4415e895e2	update	2025-11-20 12:20:46 +01:00
Dongho Kim	2034d96c9e	migration	2025-11-20 10:27:30 +01:00
Dongho Kim	5e9820136f	update	2025-11-19 13:20:56 +01:00
Dongho Kim	84fce9a82c	update	2025-11-18 14:45:41 +01:00
Dongho Kim	2e80d64ff6	update dns	2025-11-18 11:52:09 +01:00
Dongho Kim	467886a5dd	dns	2025-11-18 11:48:46 +01:00
dongho	2c1d7fa81a	update	2025-11-18 07:42:19 +00:00
Dongho Kim	c4b0e800e0	Merge pull request 'news' (#1 ) from news into master Reviewed-on: #1	2025-11-17 21:10:34 +09:00