rag-search

Semantic search on documents finds relevant information based on meaning rather than exact keywords. This Skill uses LlamaIndex and FAISS to index document collections and retrieve contextually relevant results for research and policy writing workflows.

Yes, you can combine vector search and BM25 search for document retrieval through hybrid search. This approach merges vector-based semantic matching with BM25 keyword search to improve accuracy and relevance ranking across document collections.

The best way to start running semantic search queries is using the /rag-search command followed by your query text. This triggers the hybrid retrieval pipeline to load your documents via LlamaIndex and return meaning-based matches.

Yes, you need LlamaIndex and FAISS installed to use this hybrid search approach. LlamaIndex handles document loading and indexing, FAISS provides vector storage, and BM25Retriever executes keyword search to complete the pipeline.

BM25Retriever works with FAISS vector storage for information retrieval by running parallel queries. The hybrid mechanism merges semantic vector results from FAISS with keyword-based BM25 results to optimize meaning-based relevance.

You should not use semantic search for document analysis when you need exact string matching or strict keyword filtering. Semantic search optimizes for meaning-based relevance, which may bypass exact phrases required for compliance or legal document review.