meeting-ingestion

Meeting transcript ingestion converts raw text into structured pages by creating meeting records with attendees, summaries, decisions, and action items. It enriches attendees, propagates referenced entities, and merges timeline entries across all mentioned companies and people.

Parsing meeting transcripts for action items involves extracting decisions and tasks, then running mandatory attendee enrichment to create people pages with timeline entries. This ensures every attendee becomes searchable and linked across the knowledge graph before ingestion completes.

Knowledge graph back-links enforce bidirectional connections between meeting pages and entity pages. When entity propagation creates or updates company pages, it automatically back-links them to the source meeting, treating unlinked mentions as broken graph connections.

Timeline merging propagates a single meeting event across all mentioned entities' timelines simultaneously. When ingesting meeting notes, the same event merges into each referenced company or person page, maintaining consistent dated context across the knowledge graph.

Yes, attendee enrichment must complete before meeting ingestion is considered done. The process creates or updates a people page for every attendee and adds corresponding timeline entries, ensuring all participants have dated context and searchable profiles.

Transcript parsing requires existing or newly created entity pages to establish bidirectional back-links. When referenced companies or entities lack corresponding pages, the knowledge graph treats unlinked mentions as broken connections, preventing proper timeline merging and attendee enrichment.