Reading Between the Docs: From Raw Text to Context Graph
Details
Hit the ceiling of chunk-and-embed? In this live demo, we'll load a documentation corpus into Arango AutoGraph Studio and turn it into a context graph with no hand-written schema and no manual tagging.
Watch AutoGraph map the corpus into topic domains, generate an ontology for each one, and extract entities and relationships.
Then see why some domains earn full GraphRAG while others stay vector-only. We'll finish by deploying a GraphRAG retriever and putting multi-hop questions to Deep Search.
This session will be led Pratham Batra, AI Team Lead with Arango, and he will share a demo and walk through.
What you'll learn:
- How to turn unstructured documentation into a context graph automatically
- When GraphRAG beats vector-only retrieval, and how to match the strategy to each domain
- How Deep Search answers multi-hop questions, and how to trace each answer back to its source