Semantic search ranks results by meaning rather than exact keywords, using embeddings to match paraphrases and synonyms that keyword search misses. It is the retrieval backbone of most RAG systems, and is frequently combined with keyword search in a hybrid setup.
Why it matters
Keyword search misses paraphrases and synonyms. Semantic search finds the right content by meaning, which is what users usually expect.
How it works
Both query and content are embedded, and results are ranked by vector similarity. Many systems combine semantic and keyword search (hybrid) to get both recall and exact-match precision.