Semantic search tries to understand what a query means and find related content even when the exact words are different. It often uses embeddings to compare queries and documents.
In practice
For example, a search for "refund after cancellation" might find a policy section titled "subscription termination and reimbursement" even without matching the same words.
What to watch
Semantic search can retrieve conceptually similar but incorrect material. Many systems combine semantic search with keyword filters, metadata, and reranking.