All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
- Initial open source release of VKRA Protocol
- LLMAOrchestrator: Main orchestrator coordinating all modules in the pipeline
- Presentation Module: ProductCard creation from database results
- Commission Module: Commission rate extraction from product data
- Prediction Module: Conversion probability prediction with personalization
- SR_query (Polite Ad metric) calculation
- SR_history (personalization metric) calculation
- Conversion probability prediction
- Ranking Module: Affiliate ranking algorithm balancing revenue with user satisfaction
- User Preference Module: User profile management with GDPR compliance
- Abstract Interfaces:
VectorDatabaseinterface for vector searchUserProfileStoreinterface for user profile management
- Schemas: Pydantic models for all request/response types
- Module Configuration: Request-level module customization
- Privacy Support: Three privacy levels (minimal, standard, full) with GDPR compliance
- Documentation:
- README with quick start guide
- API documentation
- Examples in multiple languages (Python, JavaScript, TypeScript, curl)
- Testing Infrastructure:
- Unit test framework with mock adapters
- Integration test structure for vertical slices
- CI/CD pipeline with GitHub Actions
- N/A (initial release)
- N/A (initial release)
- N/A (initial release)
- N/A (initial release)
- GDPR-compliant privacy handling with pseudonymization
- Differential privacy support for full privacy level
- No PII storage in vector database
- (Future features will be listed here)
- (Future changes will be listed here)
- (Future fixes will be listed here)