No Task Left Behind: Multi-Task Learning of Knowledge Tracing and Option Tracing for Better Student Assessment

Student assessment is one of the most fundamental tasks in field AI Education (AIEd). One common approach to student Knowledge Tracing (KT), which evaluates a student's knowledge state by predicting whether will answer given question correctly or not. However, context multiple choice (polytomous) questions, conventional KT approaches are limited that they only consider binary (dichotomous) correctness label (i.e., correct incorrect), and disregard specific option chosen student. Meanwhile, Option (OT) attempts model choose for question, but overlooks information. In this paper, we propose Dichotomous-Polytomous Multi-Task Learning (DP-MTL), multi-task learning framework combines OT more precise assessment. particular, show objective acts as regularization term DP-MTL framework, an appropriate architecture applying our method on top existing deep learning-based models. We experimentally confirm significantly improves both performances, also benefits downstream such Score Prediction (SP).