Hyper-graph enhanced causal recommendation with graph contrastive networks from hyperbolic space in MOOCS

Massive Open Online Courses (MOOCs) offer vast academic resources like online course but also introduce challenges such as information overload, making it difficult for users to find relevant content. Recommendation Systems (RS) aim to address this issue by predicting user preferences and suggesting suitable courses. However, traditional RS methods, such as collaborative filtering, content-based, and hybrid approaches, face limitations including data sparsity, cold-start problems, and an inability to effectively leverage graph information within MOOCs. Graph-based models have been proposed to resolve these limitations, but they often struggle with sparsity issues, leading to biased recommendations, or generate graph noise due to improper contrasting pairs. Moreover, existing methods typically overlook the role of causal inference in user performance, which is crucial for providing personalized and meaningful recommendations in MOOCs. To address these challenges, we propose the Hypergraph enhanced Causal Recommendation framework with Hyperbolic Graph Contrastive Networks (HGCR- HGCN), which leverages hyperbolic graph contrastive networks to model user-user and concept-concept relationships more effectively. Our approach introduces hypergraph representations of user-concept interactions, employs both hyperbolic and Euclidean space views of graph structures, and optimizes recommendation accuracy through contrastive learning. Additionally, causal inference is incorporated to identify key factors influencing user performance. The framework is evaluated against state-of-the-art baselines using a large-scale MOOC dataset, demonstrating improvements in recommendation accuracy, diversity, and novelty. Our findings suggest that HGCR-HGCN offers a robust solution to the problems of data sparsity, biased recommendations, and user performance analysis in MOOCs. In summary, this research presents an innovative approach that combines hypergraph modeling, contrastive learning, and causal inference to deliver more accurate and personalized recommendations in MOOCs.

Text FSKTM 2024 21 - Full Text.pdf Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (2MB) Text FSKTM 2024 21.pdf Restricted to Repository staff only Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (3MB)

Actions (login required)

View Item