For an LLM application, a knowledge base is a collection of trusted documents that can supply relevant context to answer questions. In a retrieval-augmented generation (RAG) workflow, documents are split into chunks, converted into embeddings, and stored—often in a vector database—so relevant content can be retrieved and added to a prompt. section 1.4.2 section 6.1.2
A typical RAG knowledge-base workflow:
- Load documents from sources such as PDFs, web pages, or databases. section 1.7
- Split large documents into smaller text chunks. section 1.4.2
- Embed the chunks and store the text alongside its vectors in a vector store for similarity-based retrieval. section 1.4.2 section 1.7
Enterprise question answering
A chatbot can retrieve focused passages from company documents or an intranet to answer a question. Supplying only relevant content avoids overloading the prompt with entire large documents and can improve speed, cost-efficiency, and accuracy. section 6.1.2
Figure 1.9 A collection of documents is split into text chunks and transformed into vector-based embeddings. Both text chunks and related embeddings are then stored in a vector store.
Shows documents split into text chunks, transformed into embeddings, and stored together in a vector store. section 1.4.2
