Scholarlink

<p>Knowledge tracing (KT) refers to the problem of predicting a learner’s future performance based on their past performance in education. Recently, attention-based sequence modeling methods achieve impressive predictive performance. However, existing solutions merely consider one single sequence modeling method, which might fail to capture the comprehensive state of knowledge across long sequences. In this paper, we propose <b>D</b>ual <b>S</b>equence <b>M</b>odeling for <b>K</b>nowledge <b>T</b>racing (DSMKT). DSMKT aims to enhance the modeling of a learner’s long-term profile by collaborating two sequence modeling methods, i.e.,&#xa0;the masked self-attention mechanism and the gated recurrent unit. To further exploit the synergy between two sequence models, we adopt the idea of online knowledge distillation and adaptively combine two branches to form a stronger teacher model, which in turn provides predictions as extra supervision for better modeling ability. Extensive experiments on four real-world benchmark datasets show that DSMKT performs excellently in predicting future learner responses.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Dual Sequence Modeling for Knowledge Tracing

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年6月10日

查看全文
  • Qian Ning,
  • Chengyu Guo,
  • Kunjia Liu,
  • Jiuyang Tang,
  • Shiqi Zhang,
  • Weixin Zeng,
  • Xiang Zhao

摘要

Knowledge tracing (KT) refers to the problem of predicting a learner’s future performance based on their past performance in education. Recently, attention-based sequence modeling methods achieve impressive predictive performance. However, existing solutions merely consider one single sequence modeling method, which might fail to capture the comprehensive state of knowledge across long sequences. In this paper, we propose Dual Sequence Modeling for Knowledge Tracing (DSMKT). DSMKT aims to enhance the modeling of a learner’s long-term profile by collaborating two sequence modeling methods, i.e., the masked self-attention mechanism and the gated recurrent unit. To further exploit the synergy between two sequence models, we adopt the idea of online knowledge distillation and adaptively combine two branches to form a stronger teacher model, which in turn provides predictions as extra supervision for better modeling ability. Extensive experiments on four real-world benchmark datasets show that DSMKT performs excellently in predicting future learner responses.