Data Storage
After creating the knowledge representations (graph nodes/edges and embeddings), you will need to store them in a way that allows efficient retrieval. This is where Data Storage comes in. There are two options:
-
Vector Database: A database optimized for storing embeddings (vectors) and performing similarity search quickly.
-
Graph Database: A database for the knowledge graph, optimized for traversing relationships and executing graph queries.
In a RAG setting, they represent a long-term memory based on your knowledge and document library and are a place where the preprocessed knowledge lives persistently.