Release v3.6.0

🚀 New Features

Entity Extraction & Named Entity Recognition

• NEW: Added support for entity extraction using spaCy NLP library
• NEW: Support for custom entity patterns using regex
• NEW: Configurable entity extraction with SpacyEntityExtractionConfig
• Enable with extract_entities=True in ExtractionConfig
• Results available in ExtractionResult.entities

Keyword Extraction

• NEW: Added keyword extraction capabilities using KeyBERT
• NEW: Configurable number of keywords to extract
• Enable with extract_keywords=True in ExtractionConfig
• Results available in ExtractionResult.keywords with confidence scores

Language Detection Integration

• FIXED: Language detection now properly integrated into extraction pipeline
• Automatic language detection when auto_detect_language=True
• Results available in ExtractionResult.detected_languages
• Proper error handling for missing dependencies

🔧 Improvements

• ENHANCEMENT: Replaced GLiNER with spaCy for better entity extraction performance
• ENHANCEMENT: Improved post-processing pipeline architecture
• ENHANCEMENT: Better separation of sync and async processing logic
• ENHANCEMENT: Enhanced error handling for optional dependencies

🛠️ Technical Changes

• Abstracted validation and post-processing logic into helper functions
• Improved dependency management for optional features
• Enhanced Docker workflow reliability
• Updated AI assistant configuration
• Improved exception handling for missing dependencies

📦 Dependencies

New Optional Dependencies

• spacy: For entity extraction (install with kreuzberg[entities])
• keybert: For keyword extraction (install with kreuzberg[keywords])
• fast-langdetect: For language detection (install with kreuzberg[langdetect])

Installation Examples

# Basic installation
pip install kreuzberg

# With entity extraction
pip install "kreuzberg[entities]"

# With keyword extraction  
pip install "kreuzberg[keywords]"

# With language detection
pip install "kreuzberg[langdetect]"

# All features
pip install "kreuzberg[all]"

🐛 Bug Fixes

• FIXED: Language detection now properly called during extraction
• FIXED: MissingDependencyError properly raised for missing optional dependencies
• FIXED: CI test failures related to language detection
• FIXED: Docker workflow reliability issues

💥 Breaking Changes

None - all changes are backward compatible.

📚 Documentation

• Added documentation for entity extraction configuration
• Added documentation for keyword extraction features
• Enhanced examples for new extraction capabilities

Full Changelog: v3.5.0...v3.6.0