Abstract:Massive Open Online Course (MOOC) recommendations that fail to align with the learners' prior knowledge have the potential to adversely affect educational outcomes. Despite the advancements in deep learning-based course recommendation (CR) methods, there remains a lack of comprehensive examination concerning the biases associated with the diverse knowledge backgrounds of learners. Furthermore, the phenomenon of popularity bias exists in current CR systems. In light of the above issues, this study proposes a model called Contrastive Learning and Graph Convolution Network-based Attentive Decay Network (CLGADN), which aims to improve fairness in CR by taking into account the learners' knowledge backgrounds. Specifically, (1) CLGADN employs contrastive learning to recognize the diverse knowledge backgrounds of learners and to address the challenge of popularity bias within CR, and (2) A monotonic attention decay mechanism is incorporated into the CLGADN to account for the knowledge forgetting curve, acknowledging that the knowledge learners have recently acquired shapes their understanding of the new course, more than the knowledge obtained in the past. Real-world XuetangX data are used to evaluate the proposed method. Experimental results reveal that (1) the CLGADN outperforms other recent CR methods regarding accuracy and fairness, achieving 74.73% on HR @10, 48.35% on NDCG @10, 41.45% on MRR , and 3.9% on TotalScore , a metric for evaluating whether the recommendations align with the learner's knowledge background, and (2) Multi-graph contrastive learning can improve fairness by dealing with the issues of sparse data and popularity bias. This study provides insights for MOOC platforms enhancing the fairness of CR algorithms by considering the varied knowledge backgrounds of different users. It can potentially mitigate the negative effects on learners' educational outcomes by recommending courses aligned with their knowledge backgrounds.

EduLGCL: Local-global contrastive learning model for education recommendation

Knowledge-Aware Attentive Embedding Learning for Educational Group Recommendation

Course Recommendation Model Based on Layer Dropout Graph Differential Contrastive Learning

EduGraph: Learning Path-based Hypergraph Neural Networks for MOOC Course Recommendation

Integrating learners' knowledge background to improve course recommendation fairness: A multi-graph recommendation method based on contrastive learning

Heterogeneous Graph Contrastive Learning for Recommendation

RDGCL: Reaction-Diffusion Graph Contrastive Learning for Recommendation

GNNCL: A Graph Neural Network Recommendation Model Based on Contrastive Learning

Contrastive Graph Learning for Social Recommendation

RecDCL: Dual Contrastive Learning for Recommendation

Multimodal Graph Contrastive Learning for Multimedia-Based Recommendation

LightGCL: Simple Yet Effective Graph Contrastive Learning for Recommendation

Intent-guided Heterogeneous Graph Contrastive Learning for Recommendation

Multi-behavior-based graph contrastive learning recommendation

XSimGCL: Towards Extremely Simple Graph Contrastive Learning for Recommendation

Supervised Contrastive Learning for Recommendation

Community-Enhanced Contrastive Learning for Graph Collaborative Filtering

L^2CL: Embarrassingly Simple Layer-to-Layer Contrastive Learning for Graph Collaborative Filtering

MMGCL: Meta Knowledge-Enhanced Multi-view Graph Contrastive Learning for Recommendations

Multi-knowledge enhanced graph convolution for learning resource recommendation

Enhancing Knowledge-Concept Recommendations with Heterogeneous Graph-Contrastive Learning