Data/ML/AIDatabases
Knowledge Graph / GraphRAG
Also written as Knowledge Graph, GraphRAG
Storing facts as a network of linked entities (people, companies, products and how they relate) and using it to help an AI answer questions that need connecting several facts together. GraphRAG is the name for combining a knowledge graph with RAG.
Think of it like
A detective's evidence board with string between the photos, instead of a pile of separate case notes.
Junior or senior?
Building and maintaining the graph is the hard part.
Senior sounds like
Can explain how entities were extracted and kept up to date, and why ordinary RAG wasn't enough for their questions.
Ask them
“What questions did the graph answer that normal RAG couldn't, and how did you keep the graph up to date?”