EKT-ML: An efficient knowledge tracing model with multi-task learning

doi:10.1016/j.eswa.2025.129671

EKT-ML: An efficient knowledge tracing model with multi-task learning

2025-09-11

0

PRE

AI

Prof. LI Qing (Qing Li)

DOI：10.1016/j.eswa.2025.129671

Original

Original request for help

Save

Abstract

En 中文

• A self-attention-based Knowledge Tracing (KT) model, EKT-ML, is proposed. • In EKT-ML, an MTL component is proposed to mitigate incorrect Q-KC mapping issue. • A PII component is proposed to capture and utilize the repeated attempts pattern. • An SVD-MLP component is proposed to reduce the time complexity of EKT-ML. • Experimental results show that EKT-ML outperforms the benchmark KT models.

EKT-ML: An efficient knowledge tracing model with multi-task learning

Abstract

Journal

Organization

Cited Papers