Knowledge graphs usually treat facts as fixed, but calling something a fact is a decision. Someone applied a rule, in a certain context and at a certain moment; then the customer becomes a partner, the number changes, and most systems simply overwrite the old value… Along with it goes any record of why it was ever true.
One way to address this is with “truth receipts.” These are metadata that stay attached to a fact and record the rules behind it, when it was decided and when it stopped being true. Drawing on their decade-plus experience, Himanshu Singh (Netflix), Amy Hodler (GraphGeeks) and Gerald Mann (Graphwise) dive into how it works in practice. You’ll leave knowing how to make your graph explain itself, so the next time a fact changes, you can say exactly why.
Bring your questions to the live conversation!
You will learn:
- How truth receipts track the rules, context and timing behind a fact, including when it stopped being true
- How to spot shadow concepts, where teams use a term differently from its formal definition
- How LLMs can surface terminology drift from logs and transcripts
- When to update your ontology and when to bring teams back in line