AI & LLMs
Semantic search
Semantic search matches on meaning instead of exact keywords, usually by combining embeddings with traditional text ranking.
Semantic search returns results based on what a query means, not just the words it contains. It typically embeds both the query and the documents, then ranks by vector similarity - often blended with classic keyword (full-text) scoring for the best of both.
The blend matters: keywords are precise for exact terms and names, while embeddings catch paraphrases and related concepts.
In GraphCanon
GraphCanon's search combines full-text ranking with embedding similarity, so a query like "framework for chaining prompts" can surface the right tools even without an exact name match.
See also
Related terms
Last reviewed 2026-07-09