We (Tilores) have used our entity resolution engine in a similar way for RAG, in an insurance company use case. I guess it would be the Graph-Enhanced Hybrid Retrieval or Knowledge Graph-Based Query Augmentation and Generation approach as you describe. In that case, the LLM is using our entity resolution engine as a fuzzy keyword search to find a customer profile, and what insurance policies they have, and then the vector database is returning information about the insurance policy.
We have proposed the term "EntityRAG" for this, but maybe in the end we just get lumped in with Graph databases as GraphRAG. tbc.
We (Tilores) have used our entity resolution engine in a similar way for RAG, in an insurance company use case. I guess it would be the Graph-Enhanced Hybrid Retrieval or Knowledge Graph-Based Query Augmentation and Generation approach as you describe. In that case, the LLM is using our entity resolution engine as a fuzzy keyword search to find a customer profile, and what insurance policies they have, and then the vector database is returning information about the insurance policy.
We have proposed the term "EntityRAG" for this, but maybe in the end we just get lumped in with Graph databases as GraphRAG. tbc.
You can see a video about it with Kern.ai here: https://youtu.be/QUdIpXUxaAc