Skip to Content
Adaptive LearningOverview & Courses

Adaptive Learning

Creation extends to teaching. The Create Course lens authors a structured learning module from your source video — semantic search surfaces the key moments and segments that become its sections — and the system adapts to each learner as they go.

Course generation

The Analyst builds a course from the video’s analysis:

  • Summary and key moments (found via semantic search) as orientation
  • Learning modules — text, video segments, and quizzes (src/components/course/learning-modules/)
  • Quizzes whose results feed the knowledge-tracing engine

The Digital Learner Profile (DLP)

As you take quizzes, Aura tracks performance per knowledge component using Bayesian Knowledge Tracing with temporal decay. The DLP provides calibrated probability-of-mastery estimates — visible in the DLP Monitor (src/components/analysis/DlpMonitor.tsx).

Content selection is prerequisite-aware: the Graph Knowledge service maintains a concept graph, computes learning paths, and flags prerequisite gaps before recommending what to study next.

With explicit consent (src/components/course/ConsentModal.tsx), the Biofeedback Monitor (useBiofeedback, face-api.js) can estimate engagement during study sessions to inform pacing. All processing is local; nothing leaves the browser.

Privacy-preserving improvement

Aggregate learning improves through Federated Learning with differential privacy — the system gets better across users while individual data stays private.

Project ValhallaBayesian Knowledge Tracing