Implementing Amazon Neptune Analytics GraphRAG with Existing Java Development Code

AWS showed how to build GraphRAG from existing Java/Spring code by extracting a knowledge graph schema from ORM and service logic.

The process creates a spec.json, loads it into Amazon Neptune Analytics, and uses Amazon Bedrock Knowledge Bases for natural-language graph queries.

A concert ticketing example illustrated how to derive entities, relationships, and business rules, with AI and domain experts collaborating on the model.

The approach helps turn application code into a knowledge graph that improves RAG’s ability to answer relationship-heavy questions.