Recommendation Method for MOOC Based on Graph Contrastive Learning

doi:10.19678/j.issn.1000-3428.0063646

Abstract

Abstract: With the rapid development of MOOC education platforms, the number of courses and users have increased sharply, and learners often have no way to start when studying a wide variety of courses.The application of traditional recommendation methods in MOOC recommendation has a poor recommendation effect for courses with low exposure times and insufficient robustness to noise data. A MOOC recommendation method based on graph contrastive learning is proposed to provide high-quality recommendations to learners, and a new data enhancement method is designed for the bipartite graph structure.First, data enhancement is conducted on randomly added or deleted edges of the bipartite graph of user-item interaction, and two subviews are obtained.Next, the graph convolution neural network is used to extract the node features of the original bipartite graph and two subviews, obtain the node representation of users and projects, and construct the auxiliary tasks of recommendation supervision task and contrastive learning for joint optimization. Finally, the node representations of users and projects are dot products to obtain the recommendation results.A top-K recommendation is conducted on the MOOC dataset.The experimental results show that the proposed method is better than the lightgcn model.Recall@5 and NDCG@5 are significantly improved:Recall@5 by up to 7.8% and NDCG@5 by up to 7.3%.This study demonstrates that the proposed method can effectively improve the recommendation accuracy of the model for courses with low exposure times and robustness to noise data.

Key words: MOOC recommendation, Graph Convolution Network(GCN), self-supervised learning, personalized recommendation, graph contrastive learning

摘要： 随着MOOC在线教育平台的飞速发展，课程和用户数量激增，学习者在面对种类繁多的课程时往往较难选择，传统的推荐方法在MOOC课程推荐中应用存在对曝光次数较低的课程推荐效果差和对噪声数据鲁棒性不足的问题。为给学习者提供高质量的推荐，提出一种图对比学习的MOOC推荐方法，同时针对二分图结构给出一种新的数据增强方法。对输入的用户项目交互的二分图随机添加或者删除边进行数据增强，得到两个子视图，使用图卷积神经网络对原始二分图和两个子视图进行节点特征提取得到用户和项目的节点表征，并构建推荐监督任务和对比学习的辅助任务进行联合优化，在此基础上将用户和项目的节点表征进行点积获得推荐结果。在MOOC数据集上进行Top-K推荐的实验结果表明，相较于LightGCN模型，该方法在Recall@5和NDCG@5上均有显著提升，最高分别提升7.8%和7.3%，能够有效提高模型对于曝光次数较低的课程的推荐准确性和对于噪声数据的鲁棒性。

CLC Number:

TP18

WANG Shuyan, GUO Ruihan, SUN Jiaze. Recommendation Method for MOOC Based on Graph Contrastive Learning[J]. Computer Engineering, 2023, 49(1): 57-64,72.

王曙燕, 郭睿涵, 孙家泽. 基于图对比学习的MOOC推荐方法[J]. 计算机工程, 2023, 49(1): 57-64,72.

/ / Recommend / Download Citations

URL: http://www.ecice06.com/EN/10.19678/j.issn.1000-3428.0063646

http://www.ecice06.com/EN/Y2023/V49/I1/57

Figures/Tables 12

References

[1] 王春梅, 王曙燕, 孙家泽.基于MOOC+SPOC的《C语言程序设计》混合式一流课程建设研究[J].软件导刊, 2020, 19(12):10-13. WANG C M, WANG S Y, SUN J Z.The primary exploration on the construction of C programming language based on MOOC+SPOC[J].Software Guide, 2020, 19(12):10-13.(in Chinese)
[2] 任敏.个性化推荐及其在MOOC中的应用[J].信息通信, 2019, 32(9):149-151. REN M.Personalized recommendation and application in MOOC[J].Information & Communications, 2019, 32(9):149-151.(in Chinese)
[3] 刘华玲, 马俊, 张国祥.基于深度学习的内容推荐算法研究综述[J].计算机工程, 2021, 47(7):1-12. LIU H L, MA J, ZHANG G X.Review of studies on deep learning-based content recommendation algorithms[J].Computer Engineering, 2021, 47(7):1-12.(in Chinese)
[4] CHENG H T, KOC L, HARMSEN J, et al.Wide & deep learning for recommender systems[C]//Proceedings of the 1st Workshop on Deep Learning for Recommender Systems.New York, USA:ACM Press, 2016:7-10.
[5] 杨安平, 陈松乔, 胡鹏.基于图嵌入正则化的人脸线性判别分析[J].计算机工程, 2011, 37(12):164-165. YANG A P, CHEN S Q, HU F.Face linear discriminant analysis based on graph embedding and regularization[J]. Computer Engineering, 2011, 37(12):164-165.(in Chinese)
[6] 邓爱林, 朱扬勇, 施伯乐.基于项目评分预测的协同过滤推荐算法[J].软件学报, 2003, 14(9):1621-1628. DENG A L, ZHU Y Y, SHI B L.A collaborative filtering recommendation algorithm based on item rating prediction[J].Journal of Software, 2003, 14(9):1621-1628.(in Chinese)
[7] 刘建国, 周涛, 汪秉宏.个性化推荐系统的研究进展[J].自然科学进展, 2009, 19(1):1-15. LIU J G, ZHOU T, WANG B H.Research progress of personalized recommendation system[J].Progress in Natural Science, 2009, 19(1):1-15.(in Chinese)
[8] 刘峰, 王宝亮, 邹荣宇, 等.基于随机游走的网络表示学习推荐算法[J].计算机工程, 2021, 47(9):90-96, 105. LIU F, WANG B L, ZOU R Y, et al.Recommendation algorithm using network representation learning based on random walk[J].Computer Engineering, 2021, 47(9):90-96, 105.(in Chinese)
[9] VAN DEN B R, KIPF T N, MAX W L.Graph convolutional matrix completion[EB/OL].[2021-11-20].https://arxiv.org/abs/1706.02263.
[10] WANG X, HE X N, WANG M, et al.Neural graph collaborative filtering[C]//Proceedings of the 42nd International ACM SIGIR Conference on Research and Development in Information Retrieval.New York, USA:ACM Press, 2019:165-174.
[11] HE X N, DENG K, WANG X, et al.LightGCN:simplifying and powering graph convolution network for recommendation[C]//Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval.New York, USA:ACM Press, 2020:639-648.
[12] WANG F, LIU H P.Understanding the behaviour of contrastive loss[EB/OL].[2021-11-20].https://arxiv.org/abs/2012.09740.
[13] WU J C, WANG X, FENG F L, et al.Self-supervised graph learning for recommendation[C]//Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval.New York, USA:ACM Press, 2021:726-735.
[14] DEVLIN J, CHANG M W, LEE K, et al.BERT:pre-training of deep bidirectional transformers for language understanding[EB/OL].[2021-11-20].https://arxiv.org/abs/1810.04805.
[15] FANG H C, WANG S C, ZHOU M, et al.CERT:contrastive self-supervised learning for language understanding[EB/OL].[2021-11-20].https://arxiv.org/abs/2005.12766.
[16] CHEN T, KORNBLITH S, NOROUZI M, et al.A simple framework for contrastive learning of visual representations[EB/OL].[2021-11-20].https://arxiv.org/abs/2002.05709.
[17] YAO T S, YI X Y, CHENG D Z, et al.Self-supervised learning for large-scale item recommendations[EB/OL].[2021-11-20].https://arxiv.org/abs/2007.12865.
[18] LIU Z W, CHEN Y G, LI J, et al.Contrastive self-supervised sequential recommendation with robust augmentation[EB/OL].[2021-11-20].https://arxiv.org/abs/2108.06479.
[19] YOU Y N, CHEN T L, SUI Y D, et al.Graph contrastive learning with augmentations[EB/OL].[2021-11-20].https://arxiv.org/abs/2010.13902.
[20] XU M H, WANG H, NI B B, et al.Self-supervised graph-level representation learning with local and global structure[EB/OL].[2021-11-20].https://arxiv.org/abs/2106.04113.
[21] GUTMANN M U, HYVÄRINEN A.Noise-contrastive estimation:a new estimation principle for unnormalized statistical models[J].Journal of Machine Learning Research, 2010, 9:297-304.
[22] RENDLE S, FREUDENTHALER C, GANTNER Z, et al.BPR:Bayesian personalized ranking from implicit feedback[M].Corvallis, USA:AUAI Press, 2012.
[23] ZHANG J, HAO B W, CHEN B, et al.Hierarchical reinforcement learning for course recommendation in MOOCs[J].Artificial Intelligence, 2019, 33:435-442.
[24] ZHAO W X, MU S L, HOU Y P, et al.RecBole:towards a unified, comprehensive and efficient framework for recommendation algorithms[EB/OL].[2021-08-28].https://arxiv.org/abs/2011.01731.
[25] GLOROT X, BENGIO Y.Understanding the difficulty of training deep feedforward neural networks[C]//Proceedings of International Conference on Artificial Intelligence and Statistics.Boston, USA:MIT Press, 2010:249-256.
[26] XUE H J, DAI X Y, ZHANG J B, et al.Deep matrix factorization models for recommender systems[C]//Proceedings of the 26th International Joint Conference on Artificial Intelligence.[S.1.]:AAAI Press, 2017:3203-3209.

Please choose a citation manager

Content to export