Semantic Search
A search approach that understands the meaning and intent behind a query rather than just matching keywords, powered by embeddings and vector similarity.
How It Differs from Keyword Search
Keyword search matches exact terms: searching 'automobile' won't find documents about 'cars'. Semantic search converts both queries and documents into embeddings. Since 'automobile' and 'car' have similar embeddings, semantic search finds the match.
Implementation
Documents are pre-embedded and stored in a vector database (Pinecone, Weaviate, Qdrant, pgvector). At query time, the search query is embedded and the nearest vectors are retrieved using approximate nearest neighbor algorithms (HNSW, IVF).
Hybrid Search
Best results often come from combining keyword search (BM25) with semantic search, then reranking with a cross-encoder. This captures both exact term matches and semantic similarity.