This PR implements a complete RAG evaluation system that enables developers
to measure and improve RAG quality systematically.
Features:
- Dataset and test case management
- Automated evaluation execution
- Comprehensive metrics (precision, recall, F1, MRR, hit rate)
- Configuration recommendations based on results
- 20+ REST API endpoints
- 60+ unit tests with 100% coverage
Database:
- 4 new models: EvaluationDataset, EvaluationCase, EvaluationRun, EvaluationResult
- Migrations included for all new tables
Testing:
- 60+ comprehensive unit tests
- All tests passing
- 100% service method coverage
Addresses: #11656, #11381, #7735, #10323, #6182, #5546
Breaking Changes: None
Dependencies: None (uses existing RAGFlow dependencies)
