Movie Chat

Inspiration

There is a subset of problems where it makes sense to consult a structured data source and neither vector-based RAG nor plain-text LLM context is enough. For these domains, a knowledge graph can offer a superior context.

What it does

Chat app that answers questions based on a knowledge graph. Has a pretty robust pipeline that parses user's question, leverages an LLM to create a CYPHER query out of it, validates and re-constructs the CYPHER query, executes the query against the graph DB, and then recombines the DB results with LLM to make a human-readable answer.

How we built it

Langgraph, Neo4j, python, uvicorn, pydantic, vite, js, react, OpenAI

Built With

• javascript
• langgraph
• neo4j
• pydantic
• python
• react
• uvicorn
• vite