Graph-Driven RAG AI Powered by DITA
Can generative AI benefit from structured content? What role play knowledge graphs in this? And what does all of that have to do with DITA? After they discovered the shortcomings of vector-based RAG there is a buzz in the industry around knowledge graph-driven retrieval-augmented generation (Graph RAG). This session will put the question of whether and how DITA fits into that equation. Michael Iantosca, Senior Direct of Knowledge Platforms and Engineering at Avalara, and Helmut Nagy, Chief Product Officer of Semantic Web Company will demonstrate a full-scale implementation of a graph-driven RAG based on intelligent structured content. Presented by Iantosca as graph AI theory at CIDM and in Best Practices newsletters just as generative AI blasted into the mainstream, this session will explain how DITA was used to automate the construction and maintenance of a scalable knowledge graph to drive generative AI applications along with a fully functional advanced neuro-symbolic chatbot that supports what other models lack: the ability to deterministic reasoning and inference. This isn’t your typical show-and-tell session, but an in-depth review of how the model and solution were built, the theory behind it, and how other teams can replicate the same. Several years ago, some audacious Markdown maven declared that DITA was getting “long in the tooth”. Now, at the dawn of neuro-symbolic generative AI, it turns out that DITA was indeed, the future.
Key Takeaways:
Learn about the full story of a GraphRAG integration on top of DITA:
• Learn how the model was built
• Learn how the content graph as generated
• Get an idea about the integration in a GraphRAG solution
• Learn the theory behind it
