邓伟伟1 傅凌峻1 陈 寒2* 奉国和1
(1.华南师范大学 经济与管理学院,广东 广州 510006;2.华南师范大学 教师教育学部,广东 广州 510631)
【摘 要】现有网络课程推荐的相关研究较少考虑用户对不同维度课程特征偏好,针对此不足提出融合多维度特征的网络课程推荐方法。利用多种数据分析技术获取课程描述文本、课程关系数据和课程类别属性的特征表示,基于深度自编码器融合各维度特征表示以构建课程与用户特征画像,并将其输入深度神经网络实现课程推荐。实验证明,相较于基于单一维度特征方法、基于双重维度特征方法以及传统机器学习方法,该方法在Precision@K和mAP@K评价指标上均有明显提高,对促进个性化课程推荐的实践应用具有一定的指导意义。
【关键词】网络课程推荐;多维度特征;特征融合;深度学习
An Online Course Recommendation Method Based on Multidimensional Feature Fusion
DENG Weiwei1, FU Lingjun1, CHEN Han2* and FENG Guohe1
(1. School of Economics and Management, South China Normal University,Guangzhou 510006, China;2. College of Teacher Education, South China Normal University,Guangzhou 510631, China)
Abstract: To address the problem that existing course recommendation methods rarely consider users’ preferences for multidimensional course features, this article proposed an online course recommendation method that incorporates multidimensional features. Multiple data analysis techniques were used to extract features from textual, relational, and categorical course data and used a deep autoencoder to fuse all the features for profiling courses and users. Then, it input the profiles of courses and users into a deep neural network to generate recommendations. Experiments demonstrated that this method outperformed methods based on single-dimensional and two-dimensional features, as well as traditional machine learning methods in terms of Precision@K and mAP@K. The proposed method is of great significance for improving theoretical research and practical application of personalized online education.
Keywords: online course recommendation; multidimensional data; feature fusion; deep learning
下载:
基于多维度特征融合的网络课程推荐方法构建研究.pdf