After 18 years of data governance consulting, I kept running into the same problem: the rules that define how data should look — standards, regulations, contracts, SOPs — live in unstructured documents, while the data we govern lives in structured systems. We governed the downstream but left the upstream locked in Docs.

When LLMs became capable enough, I asked a simple question: What if we could automatically extract governed, structured knowledge directly from those documents — not just search them, but truly understand them with domain context?

Over the past two years, I designed and tested an ontology-driven extraction approach across multiple industries — investment contracts, government policy reports, industrial quality inspection, and enterprise semantic enrichment. The core method: domain experts define "what to look for" through reusable ontology templates, and LLMs execute a multi-phase pipeline — from entity recognition through relationship discovery to risk identification.

Speaker: Chen Liu, CEO, DGWorkshop (Beijing) Technology Consulting Co., Ltd.

Mr. Chen Liu is a Data Governance practitioner and entrepreneur with 18 years of hands-on experience across Finance, Energy, Telecom, and Manufacturing. As founder and CEO of DGWorkshop, he has led DG&M delivery for 100+ enterprise projects in China. A DAMA-certified professional (CDMP) and recipient of DAMA International and DAMA China awards, Chen has spent the past two years exploring how Large Language Models can transform unstructured documents — policies, standards, contracts, inspection reports — into structured, governed knowledge assets. His work focuses on ontology-driven knowledge extraction pipelines that bridge the persistent gap between upstream documents and downstream data systems. He believes the next frontier of data governance lies not in more documentation, but in turning text into action, especially by AI Agents.