retrieval

Definitions

1. In the context of LLM applications, the process of fetching relevant documents or data chunks from an external knowledge base to provide as context for the model's response. Retrieval quality — measured by relevance, recall, and freshness — is often the bottleneck in RAG system performance.

   In plain English: Finding the right documents to show to an AI so it can give an informed answer, like pulling the right books off a library shelf.

   Example: We improved answer quality more by fixing retrieval with better chunking and hybrid search than by upgrading the language model.

Related Terms

• RAG
• Embedding
• grounding
• RAG Search