基于用户标签和信任关系的协同过滤推荐算法研究
|
摘要
随着互联网产业的快速发展,推荐系统已成为商业领域的研究热点,传统的仅考虑用户相似度或项目相似度的推荐算法已不能满足用户对推荐效率和推荐准确率的要求.考虑到社会好友间信任关系在推荐中的有益作用,信任关系应当成为推荐系统的考虑因素之一,文章提出一种基于标签和信任关系的协同过滤模型.首先,根据用户标签筛选出相似度较高的用户,根据他们对项目的评价预测得分;然后,根据社区内信任关系计算基于信任的评分;最后,综合两项得分进行预测.通过Epinions数据集验证表明:对比单纯的相似度推荐,添加信任因素后推荐结果有明显改变且随着信任网络规模扩大,项目预测得分趋于稳定,预测精度明显提高,更适用于移动电子商务环境下的个性化推荐问题.
With the rapid development of the Internet industry,the recommendation system has become a research hotspot in the commercial field.The traditional recommendation algorithm that only considers user similarity or project similarity cannot meet the user's requirements on recommendation efficiency and recommendation accuracy.Considering the beneficial role of trust between social friends in recommendation,trust relationship should be one of the factors to be considered in recommendation system.This paper proposes a collaborative filtering model based on tag and trust relationship.Firstly,users with high similarity were selected according to the user labels,and the score was predicted according to their evaluation of the project.Then,trust-based ratings were calculated based on trust relationships within the community.Finally,the forecast is based on a combination of two scores.Through Epinions data set verification,it is shown that compared with the simple similarity recommendation,the recommendation results have significantly changed after the addition of trust factors.With the expansion of the trust network,the project prediction score tends to be stable and the prediction accuracy is significantly improved,which is more suitable for the personalized recommendation in the mobile e-commerce environment.
With the rapid development of the Internet industry,the recommendation system has become a research hotspot in the commercial field.The traditional recommendation algorithm that only considers user similarity or project similarity cannot meet the user's requirements on recommendation efficiency and recommendation accuracy.Considering the beneficial role of trust between social friends in recommendation,trust relationship should be one of the factors to be considered in recommendation system.This paper proposes a collaborative filtering model based on tag and trust relationship.Firstly,users with high similarity were selected according to the user labels,and the score was predicted according to their evaluation of the project.Then,trust-based ratings were calculated based on trust relationships within the community.Finally,the forecast is based on a combination of two scores.Through Epinions data set verification,it is shown that compared with the simple similarity recommendation,the recommendation results have significantly changed after the addition of trust factors.With the expansion of the trust network,the project prediction score tends to be stable and the prediction accuracy is significantly improved,which is more suitable for the personalized recommendation in the mobile e-commerce environment.
引文
[1]肖会敏,张锟,崔春生.基于协同过滤的移动电子商务推荐算法.系统科学与数学,2016,36(8):1266-1274.(Xiao H M,Zhang K,Cui C S.Recommendation algorithms for mobile e-commerce based on collaborative filtering.Journal of Systems Science and Mathematical Sciences,2016,36(8):1266-1274.)
[2]陈美灵.移动环境下基于情境感知的推荐系统研究.硕士论文,杭州电子科技大学,2017.(Chen M L.Research on context aware recommendation system in mobile environment.Master's thesis,Hangzhou Dianzi University,2017.)
[3]Symeonidis P,Tiakas E,Manolopoulos Y.Product recommendation and rating prediction based on multi-modal social networks.ACM,2011,61:61-68.
[4]邓晓懿,金淳,韩庆平,等.基于情境聚类和用户评级的协同过滤推荐模型.系统工程理论与实践,2013,33(11):2945-2953.(Deng X Y,Jin C,Han Q P,et al.Improved collaborative filtering model based on context clustering and user ranking.Systems Engineering—Theory&Practice, 2013,33(11):2945-2953.)
[5]张鹏飞,王宜贵,张志军.融合标签和多元信息的个性化推荐算法研究.计算机工程与应用,1-9[2018-08-06].http://kns.cnki.net/kcms/detail/11.2127.TP.20180511.1717.014.html.(Zhang P F,Wang Y G,Zhang Z J.Research on personalized recommendation algorithm combining tag and multielement information.Computer Engineering and Applications:1-9[2019-03-12].http://kns.cnki.net/kcms/detail/11.2127.TP.20180511.1717.014.html.)
[6]Shi M,Liu J,Zhou D.A hybrid approach for automatic mashup tag recommendation.Journal of Web Engineering,2017,61(7-8):676-692.
[7]许棣华,王志坚,林巧民,等.一种基于偏好的个性化标签推荐系统.计算机应用研究,2011,28(7):2573-2575,2579.(Xu T H,Wang Z J,Lin Q M,et al.Personalized tag recommendation system based on preferences.Application Research of Computers,2011,28(7):2573-2575,2579.)
[8]Garg N,Weber I.Personalized,interactive tag recommendation for flickr.ACM,2008,61(7-8):67-74.
[9]张斌,张引,高克宁,等.融合关系与内容分析的社会标签推荐.软件学报,2012,23(3):476-488.(Zhang B,Zhang Y,Gao K N,et al.Combining relation and content analysis for social tagging recommendation.Journal of Software,2012,23(3):476-488.)
[10]杨亚东,熊庆国.基于动态标签偏好信任概率矩阵分解模型的推荐算法.计算机工程,2017,43(10):160-166.(Yang Y D,Xiong Q G.Recommendation algorithm based on dynamic label preference trust probability matrix decomposition model.Computer Engineering,2017,43(10):160-166.)
[11]尹光宇.社交网络中用户间信任度量模型研究.硕士论文,中国科学技术大学,合肥,2013.(Yin G Y.Researches on measurement model for trust between users in social networks.Master's thesis,University of Science and Technology of China,Hefei,2013.)
[12]张丰,王箭,赵燕飞,等.社交网络中一种基于社区推荐的信任模型.计算机科学,2014,41(5):168-172.(Zhang F,Wang J,Zhao Y F,et al.Trust model based on groups recommendation in social network.Computer Science,2014,41(5):168-172.)
[13]张大鹏,左孟飞,王新生.一种基于推荐偏差的多主体系统信任推荐模型.小型微型计算机系统,2014,35(8):1852-1855.(Zhang D P,Zuo M F,Wang X S.Deviation based trust recommendation model for multi-agent system.Journal of Chinese Computer Systems,2014,35(8):1852-1855.)
[14]郭均鹏,高成菊,赵昊.一种基于符号数据的群体推荐算法.系统工程学报,2015,30(1):127-134.(Guo J P,Gao C J,Zhao M H.Symbolic data analysis-based group recommendation algorithm.Journal o.f Systems Engineering,2015,30(1):127-134.)
[15]王茜,杨莉云,杨德礼.面向用户偏好的属性值评分分布协同过滤算法.系统工程学报,2010,25(4):561-568.(Wang Q,Yang L Y,Yang D L.Collaborative filtering algorithm based on rating distribution of attributes faced user preference.Journal of Systems Engineering,2010,25(4):561-568.)
[16]崔春生.基于Vague集理论的推荐系统中用户兴趣度的描述.系统工程理论实践,2017,37(3):752-760.(Cui C S.Description of user interest in recommendation system based on vague set theory.Systems Enginee ring—Theory&Practice,2017,37(3):752-760.)
[17]臧振春,惠娴斐,崔春生.基于Vague集的电子商务购物决策的模型及应用研究.系统科学与数学,2017,37(12):2375-2388.(Zang Z C,Hui X F,Cui C S.Model and application of E-commerce shopping decision-making based on vague set.Journal of Systems Science and Mathematical Sciences,2017,37(12):2375-2388.)
[18]王瑞琴,蒋云良,李一啸,等.一种基于多元社交信任的协同过滤推荐算法.计算机研究与发展,2016,53(6):1389-1399.(Wang R Q,Jiang Y L,Li Y X,et al.A collaborative filtering recommendation algorithm based on multiple social trusts.Journal of Computer Research and Development,2016,53(6):1389-1399.)
[19]侯成龙.融合社会化关系的推荐算法研究与实现.硕士论文,北京邮电大学,北京,2014.(Hou C L.Research and implementation of recommendation algorithm integrating social relations.Master's thesis,Beijing University of Posts and Telecommunications,Beijing,2014.)
[20]Kim Y A,Le M T,Lauw H W,et al.Building a web of trust without explicit trust ratings.IEEE,2008,531-536.
[2]陈美灵.移动环境下基于情境感知的推荐系统研究.硕士论文,杭州电子科技大学,2017.(Chen M L.Research on context aware recommendation system in mobile environment.Master's thesis,Hangzhou Dianzi University,2017.)
[3]Symeonidis P,Tiakas E,Manolopoulos Y.Product recommendation and rating prediction based on multi-modal social networks.ACM,2011,61:61-68.
[4]邓晓懿,金淳,韩庆平,等.基于情境聚类和用户评级的协同过滤推荐模型.系统工程理论与实践,2013,33(11):2945-2953.(Deng X Y,Jin C,Han Q P,et al.Improved collaborative filtering model based on context clustering and user ranking.Systems Engineering—Theory&Practice, 2013,33(11):2945-2953.)
[5]张鹏飞,王宜贵,张志军.融合标签和多元信息的个性化推荐算法研究.计算机工程与应用,1-9[2018-08-06].http://kns.cnki.net/kcms/detail/11.2127.TP.20180511.1717.014.html.(Zhang P F,Wang Y G,Zhang Z J.Research on personalized recommendation algorithm combining tag and multielement information.Computer Engineering and Applications:1-9[2019-03-12].http://kns.cnki.net/kcms/detail/11.2127.TP.20180511.1717.014.html.)
[6]Shi M,Liu J,Zhou D.A hybrid approach for automatic mashup tag recommendation.Journal of Web Engineering,2017,61(7-8):676-692.
[7]许棣华,王志坚,林巧民,等.一种基于偏好的个性化标签推荐系统.计算机应用研究,2011,28(7):2573-2575,2579.(Xu T H,Wang Z J,Lin Q M,et al.Personalized tag recommendation system based on preferences.Application Research of Computers,2011,28(7):2573-2575,2579.)
[8]Garg N,Weber I.Personalized,interactive tag recommendation for flickr.ACM,2008,61(7-8):67-74.
[9]张斌,张引,高克宁,等.融合关系与内容分析的社会标签推荐.软件学报,2012,23(3):476-488.(Zhang B,Zhang Y,Gao K N,et al.Combining relation and content analysis for social tagging recommendation.Journal of Software,2012,23(3):476-488.)
[10]杨亚东,熊庆国.基于动态标签偏好信任概率矩阵分解模型的推荐算法.计算机工程,2017,43(10):160-166.(Yang Y D,Xiong Q G.Recommendation algorithm based on dynamic label preference trust probability matrix decomposition model.Computer Engineering,2017,43(10):160-166.)
[11]尹光宇.社交网络中用户间信任度量模型研究.硕士论文,中国科学技术大学,合肥,2013.(Yin G Y.Researches on measurement model for trust between users in social networks.Master's thesis,University of Science and Technology of China,Hefei,2013.)
[12]张丰,王箭,赵燕飞,等.社交网络中一种基于社区推荐的信任模型.计算机科学,2014,41(5):168-172.(Zhang F,Wang J,Zhao Y F,et al.Trust model based on groups recommendation in social network.Computer Science,2014,41(5):168-172.)
[13]张大鹏,左孟飞,王新生.一种基于推荐偏差的多主体系统信任推荐模型.小型微型计算机系统,2014,35(8):1852-1855.(Zhang D P,Zuo M F,Wang X S.Deviation based trust recommendation model for multi-agent system.Journal of Chinese Computer Systems,2014,35(8):1852-1855.)
[14]郭均鹏,高成菊,赵昊.一种基于符号数据的群体推荐算法.系统工程学报,2015,30(1):127-134.(Guo J P,Gao C J,Zhao M H.Symbolic data analysis-based group recommendation algorithm.Journal o.f Systems Engineering,2015,30(1):127-134.)
[15]王茜,杨莉云,杨德礼.面向用户偏好的属性值评分分布协同过滤算法.系统工程学报,2010,25(4):561-568.(Wang Q,Yang L Y,Yang D L.Collaborative filtering algorithm based on rating distribution of attributes faced user preference.Journal of Systems Engineering,2010,25(4):561-568.)
[16]崔春生.基于Vague集理论的推荐系统中用户兴趣度的描述.系统工程理论实践,2017,37(3):752-760.(Cui C S.Description of user interest in recommendation system based on vague set theory.Systems Enginee ring—Theory&Practice,2017,37(3):752-760.)
[17]臧振春,惠娴斐,崔春生.基于Vague集的电子商务购物决策的模型及应用研究.系统科学与数学,2017,37(12):2375-2388.(Zang Z C,Hui X F,Cui C S.Model and application of E-commerce shopping decision-making based on vague set.Journal of Systems Science and Mathematical Sciences,2017,37(12):2375-2388.)
[18]王瑞琴,蒋云良,李一啸,等.一种基于多元社交信任的协同过滤推荐算法.计算机研究与发展,2016,53(6):1389-1399.(Wang R Q,Jiang Y L,Li Y X,et al.A collaborative filtering recommendation algorithm based on multiple social trusts.Journal of Computer Research and Development,2016,53(6):1389-1399.)
[19]侯成龙.融合社会化关系的推荐算法研究与实现.硕士论文,北京邮电大学,北京,2014.(Hou C L.Research and implementation of recommendation algorithm integrating social relations.Master's thesis,Beijing University of Posts and Telecommunications,Beijing,2014.)
[20]Kim Y A,Le M T,Lauw H W,et al.Building a web of trust without explicit trust ratings.IEEE,2008,531-536.