CogRec: Neural Cognitive-Style Modeling for Adaptive Learning Path Recommendation in Communication-Centric Environments

View at publisher Summarise with VIRAT
Authors: Huiying Wan, Yanan Gong
Cited by: 0
References: 18

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

In online learning platforms, intelligent recommendation algorithms are now an essential tool for boosting individualized learning experiences and increasing learning effectiveness. However, current approaches frequently ignore the cognitive distinctions across learners, which leads in recommendation outcomes that are not sufficiently flexible. This study suggests the CogRec (Cognitive Style-Aware Neural Recommendation) model as a solution to this problem. This model uses adaptive path optimization and multimodal cognitive style modeling to dynamically suggest learning paths. Eye movement signals, electroencephalogram (EEG) patterns, microexpression features, and learning behavior records are all used by CogRec to create a cognitive style vector. Additionally, it uses a self-attention technique to combine multimodal elements. At the modeling level, the model combines an extended restricted Boltzmann machine (RBM) with an attention-based heterogeneous graph neural network (A-HGNN). This approach captures the complex relationship between users and learning resources. Then, a cognitive style adaptation function is introduced to dynamically adjust the recommendation list based on task difficulty and individual processing preferences. This allows for the generation of personalized learning paths that are highly tailored to the learner's characteristics. The experimental part is verified based on two open education datasets (EdNet, MOOCCube) and a self-built multimodal dataset (CogLearn). At the same time, it is compared with four baseline methods: BPRMF, GRU4Rec, LightGCN and DGSR. Multi-dimensional indicators such as Precision@K, Recall@K, NDCG@K, ItemCoverage and Average-Popularity are used for comprehensive evaluation. The results show that CogRec outperforms the baseline model in accuracy and diversity indicators on all datasets. On the CogLearn dataset, CogRec's NDCG@10 is improved by about 36.5 % compared with DGSR, and it achieves the best performance in diversity and novelty indicators. The results of the ablation experiment further show that removing the cognitive style encoding module will result in a 21.7 % reduction in NDCG@10.

Fields of Study & Subjects

Recommender Systems and Techniques Intelligent Tutoring Systems and Adaptive Learning Online Learning and Analytics

Publication details

Journal / source
International Conference on Communication, Computing and Electronics Systems (ICCCES)
Pages
2244–2250
Publication date
21 January 2026
Language
EN