Automated Feedback Generation for Learner Modeling in Intelligent Tutoring Systems

  • Author / Creator
    Lu, Chang
  • Feedback is essential for knowledge acquisition, but there is a paucity of automated feedback generation frameworks in intelligent tutoring systems (ITSs) that facilitate and scaffold students’ learning across domains. This study introduces a novel framework for generating templated-based feedback to tackle the issue of automated feedback generation for different learner models. Specifically, it (1) implements several learner modeling algorithms including IRT, BKT, DKT-RNN, and DKT-RNN-LSTM; (2) devises and implements DKT-CI (i.e., DKT-RNN-LSTM with Contextualized Information) to estimate learners’ skill mastery states using both product data and process data; (3) compares these algorithms’ prediction accuracy, interpretability, and applicability on three datasets with various sizes extracted from different ITSs; and (4) introduces a framework to automatically generate template-based feedback on learners’ performance for the output of these learner models. Results revealed that (1) BKT and IRT outperformed DKT on smaller datasets, whereas DKT-CI outperformed other models on larger datasets; (2) for BKT, the proposed template-based feedback generation could produce KC-dependent feedback based on learner performance and expert-derived thresholds; (3) for DKT, the feedback-generation method could produce adaptive feedback for all KCs at every time step and plot individuals’ knowledge transfer, thus being more suitable for individualized formative tutoring. Implications regarding context-specific automated feedback provision for interactive digital learning systems are discussed. Findings from the present research facilitate the understanding of students’ learning behaviour and the dynamic knowledge acquisition process on different knowledge components in the ITSs and inform decision making on when and how to provide feedback in these systems.

  • Subjects / Keywords
  • Graduation date
    Fall 2021
  • Type of Item
  • Degree
    Doctor of Philosophy
  • DOI
  • License
    This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.