minutes-graph

You query people and topics across multiple meeting transcripts by building a cross-meeting entity graph from frontmatter and transcript data. This creates a structured JSON index of precomputed co-occurrences for instant relationship intelligence lookups.

A cross-meeting entity graph is used for detecting topic co-occurrences, trends, and first mentions across meeting data. It supports due diligence and relationship intelligence by providing structured, queryable associations without requiring manual transcript re-reading.

You build an entity index from meeting frontmatter and transcript data by running the bundled deterministic index builder script. This script processes your meeting schema to generate precomputed co-occurrence data for instant cross-meeting inquiries.

You can find everyone who mentioned a specific topic by querying the precomputed co-occurrence data in the entity graph. The index instantly reveals who knows about a topic and what other topics came up during those same meetings.

You need a frontmatter-based meeting schema to generate cross-meeting relationship insights. The deterministic index builder script relies on this structured frontmatter and transcript data to accurately construct the co-occurrence graph.

A limitation of using a precomputed co-occurrence graph is that lookups depend entirely on the index built by the script. You must rebuild the entity graph to reflect any new meetings or changes to the frontmatter before querying for updated insights.