Neo4j LLM Graph Builder
An open-source knowledge graph builder turns documents into Neo4j graphs, with Docker self-hosting and Ollama support for local extraction.
Neo4j LLM Graph Builder converts documents and web content into knowledge graphs for people building searchable collections or AI applications with Neo4j. It uses an LLM to identify entities, relationships, and their properties.
- Document and web imports: Build graphs from PDFs, Word documents, text files, YouTube videos, and web pages, with support for files in AWS S3 and Google Cloud Storage.
- Graph schemas: Use a custom schema or a configured schema to define the kinds of entities and relationships the model extracts.
- Graph inspection: View graphs from one or several sources together in Neo4j Bloom.
- Chat with your data: Ask questions about stored material and retrieve source metadata with responses; retrieval supports graph relationships, vector similarity, full-text search, and combinations of these methods.
You can self-host the browser application with Docker. Local deployments support Ollama for model extraction, alongside cloud providers such as OpenAI, Gemini, Diffbot, and Anthropic, plus OpenAI-compatible services. OpenAI and Diffbot are enabled by default; other providers need configuration. Ollama processes extraction requests locally, while cloud providers receive the requests sent to them.
The hosted website connects to Neo4j and uses Gemini for processing and chat. Embedding choices include Sentence Transformers, OpenAI, Gemini, and Amazon Titan. The application is open source under Apache 2.0 and requires a Neo4j database with APOC.
