Title: Predicting Advanced Mathematics Learning Behaviors and Academic Risks Through Multimodal Data Analysis

Abstract:

Identifying students who are at risk of falling behind and providing timely academic interventions remain significant hurdles in advanced mathematics instruction. This difficulty stems largely from the subject’s intricate conceptual structures and the non-linear nature of learning paths, both of which frequently impede student success. To address these issues, this research employs multimodal data analytics to establish a dynamic framework capable of predicting learning behaviors and issuing early academic warnings. The proposed model utilizes a hierarchical knowledge graph ontology, featuring adaptive edge weighting that adjusts based on problem complexity and individual student performance. By integrating heterogeneous graph attention mechanisms with temporal sequence modeling, the system effectively monitors the shifting knowledge states of learners.

Empirical evaluations conducted on multimodal datasets collected over an entire semester demonstrate that this approach successfully pinpoints high-risk students and accurately traces the propagation of errors. Furthermore, the implementation of targeted interventions has been shown to significantly enhance students’ mastery of the material while mitigating academic risks. These findings confirm that combining knowledge graph analytics with multimodal temporal modeling offers a more efficient and personalized method for delivering learning support in advanced mathematics education.

Source: arXiv Generated at: 2026-06-02 00:00:00 UTC