Search that matches the meaning of a query to the meaning of documents, so results can be relevant even when they share no keywords.
Unlike traditional lexical search engines (e.g., Elasticsearch), Semantic Search relies on Vector Embeddings to map queries and documents into a high-dimensional mathematical space. This lets the system return relevant results based on conceptual similarity. In a typical RAG pipeline, semantic search is the retrieval step.
See also Large Language Model, Fine-Tuning and Model Inference.