Enhancing Student Engagement and Classroom Interaction through Mobile Interactive Technologies

Abstract

Under the framework of educational informatization, mobile interactive technology has emerged as a promising approach to address inefficiencies in traditional classroom interaction and the problem of low student engagement. However, current practices continue to face critical challenges, including the lack of personalized strategy design and insufficient recognition of implicit low-engagement patterns. Using the building information modeling (BIM) courses as a case study, it has been observed that while mobile collaboration tools improve the efficiency of complex knowledge interaction, issues such as lecture-dominated discourse and an overabundance of low-level interactions remain prevalent. Existing research has largely focused on optimizing application functionalities or evaluating explicit participation indicators, often neglecting deeper structural characteristics such as students’ peripheral positions within interaction networks, sparse connectivity, and anomalies in interaction quality. Traditional graph models have been limited to direct connections and have failed to capture the potential for indirect collaboration through higher-order paths. Additionally, current deep learning approaches lack sufficient temporal modeling of dynamic interaction structures, resulting in strategy evaluation processes that remain empirically driven rather than data-driven. To address these limitations, a dynamic strategy evaluation method based on low-engagement graph link prediction was proposed. A heterogeneous graph model incorporating student, teacher, and resource nodes was constructed, supported by a subgraph extraction and line graph transformation algorithm to analyze multi-order indirect interaction paths. A tree-long short-term memory (LSTM) model was employed to aggregate edge features and generate embedded representations of interaction potential. Ranking loss training was utilized to address sample imbalance. This study transcends the limitations of explicit indicators by quantifying the structural characteristics of low-engagement students within the network. The proposed method provides data-driven support for the precise design of group collaboration suggestions and feedback optimization mechanisms on mobile platforms. A shift from “coarse-grained interaction” to “precision intervention” is thereby facilitated. These findings are expected to enhance the effectiveness of classroom interaction and contribute to greater equity in education.