Back to results for “Embedding Models for Retrieval”
Retrieval Augmented Generation (RAG) › Semantic Search in RAG Pipelines

How does Semantic Search in RAG Pipelines work

Beginner

Answer

The system converts a user's query into an embedding and compares it against stored document embeddings, retrieving the pieces of content that are most similar in meaning.
Real-world example Semantic search in a RAG pipeline can find a relevant refund policy document even if the user's question uses completely different wording than the document.

Common follow-ups: What is Retrieval Augmented Generation (RAG), How is Semantic Search in RAG Pipelines evaluated in practice, What tools are commonly used for Semantic Search in RAG Pipelines