This is a RAG middleware service that serves upper-layer applications.
This project is a complete set of RAG middleware microservices, providing unified retrieval-augmented generation capability support for upper-layer business applications. All modules are decoupled and independently deployable, covering document parsing, vector embedding, data loading, vector storage, reranking, large model proxy, gateway routing and knowledge ontology management full-link capabilities.
- rag-magicpdf
- embeding-service
- rag-loader
- rag-embedding
- rag-service
- rag-gateway
- rag-faiss
- rag-rerank
- rag-llm-proxy
- rag-ontology
| Module | Core Responsibility |
|---|---|
| rag-gateway | External traffic entry, gateway routing, auth, rate limit |
| rag-service | Full-link orchestration, unified business logic entry |
| rag-magicpdf | Complex PDF/document parsing, layout analysis |
| rag-loader | Text cleaning, chunk segmentation, preprocessing |
| embeding-service | Text vector embedding calculation |
| rag-faiss | Vector library storage, fast similarity retrieval |
| rag-rerank | Retrieval result secondary sorting & filtering |
| rag-llm-proxy | Unified LLM access, multi-model proxy management |
| rag-ontology | Knowledge graph, entity & relationship management |