Clip / AI Tools

Tag at ingestion, inherit down the derivation chain

From Citation Needed: Provenance for LLM-Built Knowledge Graphs — Daniel Chalef, Zep AI · ≈6:17

A cheap metadata-projection technique: tagging episodes once at ingest lets retrieval filter to trusted sources (e.g. verified clinical records) without re-deriving provenance.

What’s in it

A cheap metadata-projection technique: tagging episodes once at ingest lets retrieval filter to trusted sources (e.g. verified clinical records) without re-deriving provenance.

Watch this moment

Clip transcript

modeled on the graph. With metadata projection, we can also model classifications that span many different episodes. and facts derived from them. And so I'll give you an example here. In in the prior healthc care scenario, episodes may originate from an EHR record and have an EHR tag associate associated with them, but not all records are. And so on ingestion, we tag the episodes with the EHR tag. All subsequent entities and facts derived from the episode inherit the tag. And so if the agent wants to retrieve only facts from verified clinical sources, it's very simple to filter for the appropriate tag as we walk the graph. So one tagging action at ingestion supports evaluating the veracity of a