Standard RAG retrieves isolated document fragments, failing to understand complex connections between disparate files. **GraphRAG** combines structured Knowledge Graphs (e.g. Neo4j) with unstructured texts to map semantic connections globally across documents.

Cypher Query Integration

We parse documents into Entities (nodes) and Relationships (edges). When querying, we use an LLM to generate a Cypher query, fetching related entity loops before drafting responses.

// Fetch entities connected to target query
MATCH (e:Entity {name: "vLLM"})-[r:DEPENDS_ON]->(d:Entity)
RETURN d.name, type(r);