A semantic search engine finds results by meaning rather than exact words, so a query can match text that uses different wording.
In plain words
Every item in the collection, whether a sentence, paragraph or whole document, is turned into an embedding: a list of numbers placed in a shared space. When someone searches, their query is turned into an embedding too, and the engine returns the items whose embeddings sit nearest. That lets it cope with synonyms, abbreviations and typos that trip up keyword search.
A quick example
The Sentence Transformers docs give a simple case: a query asking how to learn Python online should also find an entry that asks the same thing with "on the web" in place of "online", even though the words differ.
Where you'll meet it
- Site and help-centre search.
- The retrieval step inside RAG systems.
Related
Sources
- DocsSemantic Search, Sentence Transformers documentation · read 29 Sept 2026
- DocsBuild a semantic search engine with LangChain, LangChain docs · read 29 Sept 2026
Checked by WhatIsAI against the sources above · 29 Sept 2026. Spotted an error? Report it.