Scholarlink

This study proposes a scaffolding system that provides adaptive hints using a new dynamic assessment probabilistic model, i.e., Sliding Hidden Markov Item Response Theory (SHMIRT). The SHMIRT optimizes the degree of forgetting past data for the prediction of a student’s performance by adjusting the student’s ability change throughout the learning process. Using the SHMIRT, the system provides hints so that the student’s correct response probability approaches 0.5 to each task even during long-term learning. We assess the causal effects of the proposed scaffolding mechanism using Inverse Probability Weighting (IPW) for long-term learning data in actual classes. The results demonstrate that the proposed system is effective.

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

Probability-Based Scaffolding System Using Sliding Hidden Markov IRT for Longitudinal Learning

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Maomi Ueno,
  • Yoshimitsu Miyazawa,
  • Emiko Tsutsumi

摘要

This study proposes a scaffolding system that provides adaptive hints using a new dynamic assessment probabilistic model, i.e., Sliding Hidden Markov Item Response Theory (SHMIRT). The SHMIRT optimizes the degree of forgetting past data for the prediction of a student’s performance by adjusting the student’s ability change throughout the learning process. Using the SHMIRT, the system provides hints so that the student’s correct response probability approaches 0.5 to each task even during long-term learning. We assess the causal effects of the proposed scaffolding mechanism using Inverse Probability Weighting (IPW) for long-term learning data in actual classes. The results demonstrate that the proposed system is effective.