.png){kind=icon}
.png){kind=icon}
Return
Modeling situational interest: A multimodal representation framework for intelligent assessment in learning environment
.png){kind=icon}
PRE
AI
J
Y
DOI：10.1080/15391523.2026.2643172.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Situational Interest (SI), a transient psychological state pivotal for learning engagement, remains challenging to assess due to its dynamic and multidimensional nature. This study proposes a multimodal representation framework to address this gap by integrating cognitive-affective components and multi-source theories. The framework systematically models SI through three observable dimensions: attention (quantified via head posture analysis), learning emotion (captured through facial expression recognition), and cognitive engagement (evaluated via classroom interaction metrics). A hybrid methodology combining the Delphi Method and Analytic Hierarchy Process (AHP) was employed to assign dimension-specific weights (attention: 0.47, emotion: 0.38, engagement: 0.15), balancing expert consensus with quantitative rigor. Empirical validation in K-12 science classrooms utilized advanced techniques (e.g. FSA-Net, DeepFace) to analyze synchronized multimodal data. Results demonstrated the framework’s capability to dynamically track SI fluctuations, revealing correlations between situational elements and SI trajectories. The study advances theoretical understanding of SI’s structural mechanisms and offers educators a data-driven tool for optimizing instructional strategies. This research bridges theoretical models with intelligent analytics, establishing a foundation for real-time, adaptive SI assessment in learning environments.
Keywords:
Situational Interest
multimodal representation
intelligent assessment
interest Modeling
Journal
IF:
5
Papers:
164
Citations:
2.7K
Organization
Cited Papers