Structure-aware hybrid RAG for technical document QA with KnowledgeNode trees, BM25 + dense retrieval, cross-encoder reranking, parent-section reconstruction, bounded multi-hop retrieval, explainability, entity graphs, and local/cloud execution via Foundry Local and Azure.
bm25 rag vector-search entitygraph azure-service azure-openai retrieval-augmented-generation rag-evaluation rag-pipeline rag-chatbot foundrylocal
-
Updated
Aug 6, 2026 - Python