Platform: New search features published

After months of work on the automated analysis and indexing of speeches, we can finally present the new search features! Auto-complete now includes suggested terms as well as entities that have been detected in speeches.

Behind the scenes, we have been rethinking and rebuilding the entire data import and indexing workflow from the ground up for many months. When a speech is imported, the system now automatically collects statistics on word frequencies and detected entities. This allows us to suggest terms in the auto-suggestion, including the number of times a term appears across all speeches on the platform.

These statistical analyses are already designed so that, later on, they can support more detailed linguistic evaluations by speaker, parliamentary group, and time period—but we’re not there yet.

Beyond auto-complete, we also improved the user interface in many small places and added new functionality.

For example, electoral periods are now shown in the timeline, and results can be sorted by the duration of a speech. We also made many small improvements on the profile pages for people, organisations, documents, and terms, with a particular focus on better display on mobile devices.