DIKT: A Diagnostic and Interpretable Knowledge Tracing Framework for Prediction and Model-based Learning Analysis
Organizations: Faculty of Artificial Intelligence in Education, Central China Normal University, Wuhan, 430079, China · School of Mathematics and Systems Science, Wuhan University of Science and Technology, Wuhan, 430065, China · School of Physics and Mechanics, Wuhan University of Science and Technology, Wuhan, 430065, China
Abstract
Knowledge tracing is a key component of intelligent educational systems. However, most existing methods mainly focus on response prediction, providing limited support for explaining predicted outcomes and improving students' learning states. This paper proposes Diagnostic and Interpretable Knowledge Tracing (DIKT), which integrates response prediction, skill-level diagnosis, and model-based learning analysis. DIKT generates bounded cognitive-state trajectories and decomposes the final response probability into skill-wise probabilities. The trajectories reveal students' evolving strengths and weaknesses, while the skill-wise probabilities identify the skills that most strongly constrain the prediction. Based on these diagnostic results, DIKT supports direct cognitive-state intervention and effective-learning-duration-based what-if analysis to examine how hypothetical adjustments affect current predictions and subsequent state evolution. DIKT combines Bayesian multi-head attention, conditional variational recurrent inference, a logistic-growth decoder, and multiplicative skill-wise prediction. Experiments on three benchmark datasets show that DIKT outperforms all evaluated baselines, with an average absolute AUC gain of 2.04 percentage points. Other experiments further demonstrate that DIKT supports educationally meaningful diagnostic analysis and provides informative diagnostic indicators for downstream applications, extending knowledge tracing from response prediction to interpretable diagnosis and model-based learning analysis.
Figures & tables
| Notation | Description |
| Exercise attempted at time step | |
| Binary response label at time step | |
| Skill set associated with exercise | |
| Number of knowledge skills | |
| Maximum number of skill slots for one exercise | |
| Interaction representation at time step |
| Dataset | Students | Exercises | Skills | Interactions | Avg. Length |
| ASSIST09 | 4,217 | 26,688 | 123 | 346,860 | 82.25 |
| ASSIST17 | 1,709 | 3,162 | 102 | 942,816 | 551.68 |
| Algebra2005 | 574 | 210,710 | 112 | 809,694 | 1410.62 |
| Hyperparameter | Value |
| Hidden dimension | 64 |
| Number of attention heads | 2 |
| Number of attention layers | 2 |
| Latent dimension | 16 |
| Dropout rate | 0.1 |
| Learning rate | 0.001 |
| Methods | ASSIST09 | ASSIST17 | Algebra2005 | |||
| AUC | ACC | AUC | ACC | AUC | ACC | |
| BKT | 0.6511 | 0.6342 | 0.6017 | 0.5874 | 0.6500 | 0.7553 |
| DKT | 0.7561 | 0.7362 | 0.7178 | 0.7098 | 0.8149 | 0.8097 |
| SAINT | 0.7441 | 0.7358 | 0.7087 | 0.6922 | 0.7775 | 0.7791 |
| QIKT | 0.7801 | 0.7377 | 0.7645 | 0.6985 | 0.8408 | 0.8222 |
| DSN-KT | 0.8016 | 0.7546 | 0.6886 | 0.6811 | 0.7793 | 0.8354 |
| Methods | ASSIST09 | ASSIST17 | Algebra2005 | |||
| AUC | ACC | AUC | ACC | AUC | ACC | |
| DIKT-NM | 0.7644 | 0.7406 | 0.7033 | 0.6841 | 0.7382 | 0.8349 |
| DIKT-DA | 0.7169 | 0.7665 | 0.7860 | 0.7279 | 0.7984 | 0.8421 |
| DIKT-NC | 0.6860 | 0.7613 | 0.7545 | 0.7178 | 0.7934 | 0.8393 |
| DIKT-NA | 0.6291 | 0.7433 | 0.7478 | 0.7142 | 0.7777 | 0.8375 |
| DIKT-NMo | 0.6781 | 0.7615 | 0.7546 | 0.7185 | 0.7893 | 0.8419 |
Knowledge Tracing
Knowledge tracing models how learners’ mastery of skills changes over time from their interactions with educational tasks.
See all 35 papers