Get a Demo ↗

Content Readers Can Actually Find

A vast content archive only creates value if readers can find the right thing. Thomson Reuters governs the taxonomies behind classification, so content surfaces precisely and keeps readers engaged.

Challenge

A huge, growing body of editorial content is only valuable if it can be found. Maintaining the deep, multi-level taxonomies that classification depends on was complex, and inconsistency buried good content, hurting discovery and engagement.

Solution

Thomson Reuters built a governed context layer for its multi-level editorial taxonomies. Content is classified progressively and consistently, so readers and systems surface the right material quickly and stay engaged.

Technical approach

  • One connected model of multi-level editorial taxonomies for progressive classification
  • AI-drafted, expert-approved, and kept governed and current

Results

  • Content discovery up 20-40%, from consistent classification
  • Higher reader engagement, as the right material surfaces faster
  • Taxonomy upkeep cut, with governance in one place

Thomson Reuters moves trillions of dollars across more than 210 countries, and that scale depends on data every team can trust. As the business grew, the same critical concepts carried different definitions in every system, and that fragmentation became a barrier to both accurate reporting and enterprise AI.

Fragmented reference data couldn't keep up with AI

Across payments, risk, and marketing, reference data lived in dozens of disconnected platforms. Definitions drifted, reports disagreed, and audits stretched into quarters. Every new AI initiative stalled at the same place, because no model could rely on the meaning of the underlying data. The team needed enterprise data governance that could turn fragmented metadata into one trusted, machine-readable source of truth.

"We had the data. What we didn't have was a shared, governed definition of what any of it actually meant."

One governed knowledge graph for the whole enterprise

With TopBraid EDG, Thomson Reuters modeled its core business concepts as a governed knowledge graph, mapped fragmented reference data into shared taxonomies and ontologies, and encoded data policies as code. SHACL validation kept the model consistent as it scaled, while 200+ connectors pulled metadata from across the enterprise into a single catalog. Because everything is built on open W3C standards (RDF, OWL, SHACL, and SPARQL) the foundation stays transparent, portable, and free of vendor lock-in.

"For the first time, every team reads from the same governed definitions, and our AI can explain every answer it gives."

Trusted, AI-ready data at scale

Today, governed definitions flow into every downstream system, lineage is audit-ready across the business, and AI applications draw on context they can explain, grounded answers, not guesses. Work that once took quarters of manual reconciliation now happens in weeks, and the semantic foundation keeps compounding in value as new data sources and new AI use cases come online.