Hybrid search
Combining lexical search and vector search into a shared result list.
How are word matches and similar content combined?
Hybrid search commonly combines lexical search with vector search. Both paths supply candidates that are merged into a shared ranking. Their methods can use different scoring scales, so raw scores cannot simply be averaged. Fusion can, for example, consider rank positions. Access boundaries and business filters must remain effective in both search paths.
In a fictional knowledge portal, a query reads, “M-47 jerks on startup.” The equipment identifier can support lexical search, while vector search could find differently worded descriptions of similar symptoms. If results must apply to this exact equipment, that is defined as a separate filter rule. A high similarity score does not perform that check. A planned assessment defines suitable questions and relevant documents, then compares lexical, vector and combined search for missing or unsuitable results.
Sources
Related terms
- EmbeddingAn embedding represents content as a numeric vector. The model and comparison method determine which similarities become visible.
- Vector databaseA vector database stores and searches vector representations. A similar match is not automatically a valid or permitted answer source.
- Retrieval-augmented generationRAG combines information retrieval with generated answers. Access to sources does not remove the need for quality and permission design.
- LLM evaluationLLM evaluation tests model outputs against defined tasks and criteria. One good example does not establish overall quality.
- Semantic searchSearch that finds and ranks results by their semantic relevance to the query.
- Document chunkingSplitting a document into smaller coherent sections for search and model context.
Your project with ASES
Describe your project to us. We will get back to you with an initial assessment.