A vector space is the shared mathematical space where embeddings of documents and queries can be compared. In retrieval-augmented generation, an embedding model maps both into this space so similarity search can surface relevant documents. section 4.0
Figure 4.1 Document embedding in RAG systems
Shows how document text is embedded as vectors for retrieval. section 4.0
Searching health-care policies
A policy assistant embeds policy chunks and a question such as “Is physical therapy covered?” using the same model. Pinecone compares the query vector with stored vectors and returns the closest matches, optionally applying metadata filters. section 4.0 section 4.5.4
