Enhancing Retrieval-Augmented Generation With GRIX
summary
GRIX (Graph Index) converts large document corpora into a Neo4j knowledge graph by using an LLM to populate a schema (entities, descriptions, keywords) then building nodes and relationships with Cypher; retrieval is a pipeline of search, rerank, expand (neighborhoods or shortest paths) and relies on GDs centralities and community algorithms to surface multihop or global evidence. It supports node deduplication strategies and evaluates end-to-end performance by injecting synthetic needle-in-haystack evaluation nodes and using RAGas synthetic QA, claiming improved retrieval accuracy for complex queries on Neo4j.
AI-written summary · about
chapters (7)
- 0:00Introduction to GRIX
- 3:20GRIX structure and workflow
- 5:29Knowledge graph extraction
- 7:00Information retrieval process
- 10:26Neo4j and retrieval techniques
- 11:30Visualizing the knowledge graph
- 12:14Evaluating GRIX performance
related talks
16mEnhancing Retrieval-Augmented Generation with GraphRAG Patterns in Neo4j
12mAgentic GraphRAG: Autonomous Knowledge Graph Construction and Adaptive Retrieval
19mBeyond Retrieval: Adding a Reasoning Layer to RAG
27mAdaptive Knowledge Graphs for Dynamic RAG Systems
01h 17mGraph Database and Retrieval Augmented Generation
53mPractical GraphRAG - Making LLMs smarter with Knowledge Graphs
21mAgentic GraphRAG: Autonomous Knowledge Graph Construction and Adaptive Retrieval
32mGraphRAG: Powering Up LLMs with Knowledge Graphs