###
工程科学与技术:2020,52(5):201-208
←前一篇   |   后一篇→
本文二维码信息
码上扫一扫!
基于社会信任正则化的排名推荐算法
(1.江苏理工学院 计算机工程学院,江苏 常州 213001;2.北京邮电大学 可信分布式计算与服务教育部重点实验室,北京 100876)
Ranking Recommendation Algorithm Based on Social Trust Regularization
(1.School of Computer Eng., Jiangsu Univ. of Technol., Changzhou 213001, China;2.Key Lab. of Trustworthy Distributed Computing and Service (Ministry of Education), Beijing Univ. of Posts and Telecommunications, Beijing 100876, China)
摘要
图/表
参考文献
相似文献
本文已被:浏览 1221次   下载 380
投稿时间:2019-06-26    修订日期:2020-08-31
中文摘要: 随着在线商品交易额逐年增大和社交网络不断深入发展,推荐系统已成为解决信息过载的重要工具之一。当评分矩阵数据稀疏性较大时推荐精度就会显著下降,特别是用户冷启动时该问题更加明显。因此,本文提出一种新的基于隐式反馈信息的社会化排序推荐算法。该算法首先利用矩阵分解方法计算不同项目间的用户偏好。其次,将用户偏好信息融入贝叶斯个性化排名(Bayesian personalized ranking,BPR)算法。然后,挖掘用户之间的相似关系以及信任用户的直接和间接关系,并量化用户之间的信任关系,从而研究不同项目之间用户的偏好差异。最后,将以上信任关系和BPR算法进行融合,进而构建出社会化排序推荐模型。为了验证所提出的社会化排序推荐算法,在DouBan数据集和FilmTrust数据集上,进行算法的有效性验证。通过Precision、MAP和NGCD这3种排序评估指标分别在全数据集和用户冷启动中验证本文算法与SBPR、TBPR、BPR和MostPopular等算法之间排序推荐的优劣性。实验结果表明本文算法明显优于其他对比的排序推荐算法,并可以获得更好的推荐准确率。可见本文算法可以有效改善由于数据稀疏性和用户冷启动所引起的推荐效果差的问题。
Abstract:With the increase of online commodity transaction volume and the further development of social network, recommendation system has become one of important tools to solve information overload. However, when the data sparsity of the score matrix is relatively large, the recommendation accuracy will decrease significantly, especially when the user starts cold. A new social ranking recommendation algorithm based on implicit feedback information was proposed in this paper. The user preferences among different items using a matrix factorization method were first calculated by the proposed algorithm. Secondly, the user preference information was incorporated into the Bayesian Personalized Ranking (BPR) algorithm. Then, the similar relationships between users as well as direct and indirect relationships between trusting users were mined and quantified in order to study the differences of user preferences across projects. Finally, these trust relationships were fused with the BPR algorithm to build a social ranking recommendation model. To validate the proposed social ranking recommendation algorithm, the validity of the algorithm was verified on the DouBan dataset and FilmTrust dataset. The ranking evaluation metrics of Precision, MAP, and NGCD were used to verify the merits of ranking recommendation between the proposed algorithm and SBPR, TBPR, BPR, and MostPopular algorithms. The ranking recommendation tests were performed on these two specific social datasets for the full dataset and user cold start, respectively. The experimental results demonstrated that the proposed algorithm significantly outperforms other ranking recommendation algorithms and achieves better recommendation accuracy. It can be seen that the algorithm can improve the poor recommendation performance caused by data sparsity and user cold start.
文章编号:201900637     中图分类号:TP181    文献标志码:
基金项目:国家重点研发计划项目(2017YFC1307705);江苏理工学院博士科研启动基金项目(KYY19042)
作者简介:张俐(1977-),男,副教授,博士.研究方向:机器学习和推荐系统.E-mail:zhangli_3913@163.com
引用文本:
张俐.基于社会信任正则化的排名推荐算法[J].工程科学与技术,2020,52(5):201-208.
ZHANG Li.Ranking Recommendation Algorithm Based on Social Trust Regularization[J].Advanced Engineering Sciences,2020,52(5):201-208.