“Building Knowledge Graphs with AI Agents”
There are broadly two ways to get a knowledge graph out of unstructured documents: let the LLM work out the structure as it goes, or hand it one upfront. This talk is about the second, told by someone who had never touched an ontology before being told to use one.
I’ll walk through what it looks like in practice when the schema is given to you: what the ontology and SHACL actually buy you, how AI agents do the extraction inside those constraints, and where the whole approach turned out to help less than I expected. No semantic web background needed.
Bio
Senior AI Engineer at DataArt specializing in Generative AI and LLM applications. He has worked across the full LLM stack, from multi-agent architectures and RAG systems, through structured extraction pipelines running at scale, to deploying and monitoring them in production.