Semantic search
Also: dense retrieval, vector search
Queries and documents are embedded into the same vector space and ranked by cosine similarity, usually through an approximate nearest-neighbor index. It handles synonyms and paraphrases well but is weak on exact rare tokens such as codes and IDs.
Example
The query "how do I fix a flat tire" retrieves "repairing a punctured bicycle wheel" even though the two share no words, because their embeddings point in nearly the same direction.