EngineersofAI LogoCoursesBlogResearch LabAI LettersThe LabCode BankInteractive 3DKodrEarnestJobs
Skip to main content
Interactive 3D/Graph RAG - Knowledge Graph Retrieval
Entity Types
person
org
concept
event
Query match
createdpioneeredleadsbased ontrained withcreatedbased oncore ofpublished atownscreatedbuiltbased onusesOpenAIGPT-4RLHFSam Altm…DeepMindGeminiAttentionTransfor…NeurIPS …
Vector RAG
Returns top-k semantically similar chunks. Misses relationship paths between entities - "who owns what" is invisible.
Missing: entity relationships, multi-hop paths
Graph RAG
Traverses the knowledge graph from query entities, following edges to retrieve connected facts. Multi-hop reasoning is natural.
Captures: multi-hop facts, entity context, communities
Controls
Query
Community Granularity
Clusters shown2
Entity Types
Graph RAG builds a knowledge graph from documents, then retrieves by traversing entity relationships. Colored nodes show entity types. Blue highlights show the subgraph matched by the current query. Communities (colored regions) are detected automatically.

Graph RAG - Knowledge Graph Retrieval - Interactive Visualization

Graph RAG builds a knowledge graph from documents by extracting named entities and their relationships, then applies community detection to identify clusters of related concepts. When answering a query, it traverses the graph from query-relevant entities following edges to retrieve connected facts. This enables multi-hop reasoning that pure vector similarity search cannot handle.

  • Entity extraction identifies people, organizations, concepts, and events from raw text
  • Relation edges (created, leads, based on) capture the structured knowledge vector search misses
  • Community detection clusters related entities so global summaries can answer broad questions
  • Subgraph retrieval follows paths from query entities to retrieve multi-hop contextual facts

Part of the EngineersOfAI Interactive 3D - free interactive visualizations covering every major concept in machine learning and AI engineering. Hover any element for a plain-English explanation. No code required.