Source-aware question study
Organise historical papers, topics, dates and exam contexts before identifying repeat patterns.
Our research direction brings together assessment design, previous-year papers, learning behaviour and educator judgement.
A learner’s wrong answer is a useful observation, not a diagnosis on its own. A previous-year question is a source to study, not a guarantee that the next exam will look the same. Our approach begins with those distinctions.
We are investigating a practical AI system for competitive-exam preparation and teaching support: a system capable of spotting meaningful patterns, preparing explainable drafts and suggesting next steps that people can evaluate.
Organise historical papers, topics, dates and exam contexts before identifying repeat patterns.
Look for recurrence, concept clusters, question structures and changes across paper sets.
Create candidate questions and explanations with traceable assumptions and explicit review steps.
Compare multiple attempts before proposing a practice priority or revision plan.
Educators review instructional output; learners retain agency over how they study.
Research status: AbhyasDhara Intelligence Lab AI capabilities described on this page are under development. The interactive illustrations are conceptual and do not represent live accuracy, validated model performance or current customer data.
Past papers can reveal useful tendencies, but future questions and recruitment requirements are never guaranteed. We want a model that communicates uncertainty rather than creating false confidence.
The system is intended to support educators with reviewable material and learners with actionable practice—not replace teachers, official exam notices or verified source material.
Question generation without educational judgement is not enough. The intended workflow makes review, editing and feedback first-class steps.
Explore educator research ↗A system of student signals, curriculum context and educator judgment.