Hansard Semantic Map

A deterministic map of what was discussed, when it was discussed, and who spoke. Existing BGE-M3 speech vectors are grouped into agenda-topic themes; semantic proximity is not a claim that speakers agree.

222 meetings 1,919 agenda topics 57,255 speeches 24 themes 2018–2026

Map detailTopics

Themes Years Speakers Agenda topics

How to read this map

Select a node for its evidence and use the detail slider to move from themes to individual agenda topics.

Large coloured nodes are semantic themes. White year nodes link discussion to time. Named people link to the themes in which they contributed most words. Small coloured nodes are representative agenda topics and open the exact Hansard speech used as their entry point.

Themes by spoken-word volume

Generated 2026-08-30T14:42:36Z from immutable release . The map reuses the validated BAAI/bge-m3 serving index; it runs no LLM and embeds no full transcript or local filesystem path.