Scholarlink

Learning Management Systems (LMS) are an integral part of Higher Education Institutions (HEIs). Despite the interest in modeling LMS interactions with big data, the analysis process is often neglected due to its complexity and tedious nature. As an alternative, understanding LMS analytics using graph-based representations has lately gained traction. In this vein, we present an approach that employs graph representations of students-based shared knowledge developed and predicts their quality of interaction (QoI) with the LMS. Graph Convolutional Neural (GCN) networks were applied to data collected from LMS BlackBoard at Khalifa University, UAE, from 2020 to 2023 academic years. The best performance overall was for sophomores in 2022–2023, with a Mean Square Error (MSE) of 0.0184 and a Mean Absolute Error (MAE) of 0.0995. The GCNs are projected at the scales of Colleges and Departments, showing the formation of distinct communities. This GCN-based predictive modeling is the first step in the implementation of QoI predictive knowledge-based graphs that could be used to improve the learner’s pedagogical experience.

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

Knowledge-Based Graph Representation and Prediction of Learners’ Quality of Interaction with Learning Management Systems at Higher Education Institutions

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Abdulrahman Awad,
  • Sofia B. Dias,
  • Sofia J. Hadjileontiadou,
  • Leontios J. Hadjileontiadis

摘要

Learning Management Systems (LMS) are an integral part of Higher Education Institutions (HEIs). Despite the interest in modeling LMS interactions with big data, the analysis process is often neglected due to its complexity and tedious nature. As an alternative, understanding LMS analytics using graph-based representations has lately gained traction. In this vein, we present an approach that employs graph representations of students-based shared knowledge developed and predicts their quality of interaction (QoI) with the LMS. Graph Convolutional Neural (GCN) networks were applied to data collected from LMS BlackBoard at Khalifa University, UAE, from 2020 to 2023 academic years. The best performance overall was for sophomores in 2022–2023, with a Mean Square Error (MSE) of 0.0184 and a Mean Absolute Error (MAE) of 0.0995. The GCNs are projected at the scales of Colleges and Departments, showing the formation of distinct communities. This GCN-based predictive modeling is the first step in the implementation of QoI predictive knowledge-based graphs that could be used to improve the learner’s pedagogical experience.