Research Foundation

Sudar is built on established findings from learning science, cognitive science, and adaptive learning research — from replicated classics to 2020s AI tutoring trials — with each principle mapped to a shipped feature.

The gap Sudar fills

Most LMSs serve static content: one course for all, no memory of the learner, no adaptation of sequence or difficulty. Research shows that adaptive instruction and intelligent tutoring outperform one-size-fits-all delivery. Mainstream products still lack a longitudinal learner model and memory-aware tutoring. Sudar provides learner memory, adaptive sequencing, and AI tutoring in a single open platform.

Evidence mapped to Sudar

The homepage research cards use three tiers: durable foundations, modern meta-analyses and replication studies, and recent AI-era work on personalised multimodal delivery and memory-aware tutoring.

  • Foundation · 1885

    Forgetting Curve

    Hermann Ebbinghaus · Über das Gedächtnis

    Sudar: Spaced repetition in Flashcards & Adaptive Sequencing

  • Foundation · 2009

    Multimedia Learning

    Richard E. Mayer · Multimedia Learning, 2nd ed.

    Sudar: Seven adaptive modalities per course

  • Modern validation · 2006

    Testing Effect

    Roediger & Karpicke · Psychological Science, 17(3)

    Sudar: In-module quizzes, flashcards, and struggle signals in the Digital Learner Twin

  • Modern validation · 2011

    Intelligent Tutoring Systems

    Kurt VanLehn · Educational Psychologist, 46(4)

    Sudar: Adaptive sequencing, Next Best Action, and the AI tutor sidebar

  • AI era · 2025

    AI-Augmented Textbooks

    LearnLM Team (Google) · arXiv:2509.13348

    Sudar: Author once, deliver in text, video, audio, mind map, flashcards, and more

  • AI era · 2026

    Memory-Aware AI Tutoring

    Liu et al. (AgentTutor) · arXiv:2601.04219

    Sudar: Cross-session tutor memory in ai_tutor_context and consent-governed learner model updates

Core principles (evidence base)

  • Personalisation & adaptive instruction: Learner profiles, next-best-action, adaptive path ordering, personalised welcome messages.
  • Multimodal learning: Content authored once, delivered in text, video, audio, mindmaps, flashcards, and game-based modalities.
  • Metacognition & self-regulated learning: Progress visibility, Sudar recommends, upcoming deadlines, required-path surfacing.
  • Formative assessment & retrieval practice: In-module quizzes with immediate feedback, struggle detection feeding the learner model.
  • Intelligent tutoring & dialogue: Reactive Q&A (RAG over course content), longitudinal memory, contextual help, proactive nudges.
  • Longitudinal learner model (Digital Learner Twin): Persistent learner_profiles with ai_tutor_context, next_best_action, onboarding data.
  • Learning paths & prerequisite structure: Mandatory and optional courses, unlock rules, adaptive path ordering, certifications.
  • Organisational learning & compliance: Assignments, due dates, compliance views, certificates with shareable verification.

Sudar / LAMP paper (2026)

The reference platform, Adaptive Learning Layer (ALP) plugin architecture, and economic analysis are described in our primary paper. See Research Papers for the abstract and citation.

Alignment with broader themes

Differentiation (adaptive paths, modality choice), scaffolding (Sudar as tutor), feedback (quizzes, progress, next-best-action), motivation (streaks, certificates), and accessibility (multimodal delivery, structure for assistive tech).

Open science and reproducibility

Sudar is open source. Researchers and practitioners can inspect, extend, and evaluate the implementation. Schema and event model are documented; learning_events and ai_interactions support research on engagement and tutor usage. We encourage citing the repository and this foundation when Sudar is used in studies or derivative work.

Research Papers →Mission & Vision →