Poster: Beyond Search: Putting a Knowledge Graph to Work

Drug discovery data is scattered. Public databases and internal results differ in quality and structure, so each new question means more manual data work, and AI tools lack the context they need.

Ardigen builds knowledge graphs that close this gap. We merge public and proprietary data into one graph, shaped around each client’s disease area and questions, ready for analytics, machine learning and AI agents.

ArdKG shows this approach at work. Following the PrimeKG schema1, it links ~140,000 entities of 10 types. Its chat app offers four ways to ask: plain-English questions, target ranking, link prediction (MRR 0.30 on held-out data) and a literature agent whose every citation is checked against the papers it actually found. The result: testable hypotheses for target discovery and indication expansion.