minutes-graph

To query cross-meeting co-occurrence relationships, you build a persistent entity graph index from meeting frontmatter. This structured index surfaces people-to-people and people-to-topics relationships across all meetings without re-reading transcripts.

Analyzing cross-meeting trends from frontmatter involves creating a canonical index of people, topics, and dates. By precomputing this entity graph, you can instantly query first mentions and relationship trends across all meeting minutes.

You can extract company and product entities from meeting minutes using an optional deep extraction path. This opt-in feature enables external entity discovery when your cross-meeting queries require identifying specific companies or products.

Building a cross-meeting entity graph requires the bundled graph_build.py script and access to minutes CLI data. You must execute this script against your meetings directory to construct and refresh the persistent index.

A frontmatter entity graph provides structured cross-meeting insights that traditional full-text search misses. It maps precomputed co-occurrence data and relationships, allowing fast queries across people and topics without scanning raw transcripts.

You should refresh the cross-meeting entity graph index incrementally when new meeting minutes are added. The graph_build.py script updates the persistent frontmatter index to ensure accurate cross-entity queries and trend analysis.