Selector and explainability — overview

Once student state is modeled with BKT we need two things:

1. Selector — choose a task that fits now (zone of proximal development, $P(\text{solve}) \approx 0.7$).
2. Explanation — one clear sentence for the teacher on why this task.

This section unfolds alongside the first. After “Knowledge as probability” you can jump here and watch state turn into action.

Roadmap

• Selector in action — recommend() from web/lib/bkt.ts.
• Explainability — deterministic Estonian templates.
• Class heatmap — 22×8 grid plus live widget.

Live demo

The same selector, run on the real matrix (9 micro-skills, 20 tasks, «Defineerimine» topic), lives on Progression matrix → Simulator. You can switch virtual students and watch the top-3 candidates change.