基于用户隐式行为特征的最大熵推荐算法
|
摘要
在电商领域的推荐中,由于用户购买频率较低且很少留下评价信息,使推荐系统面临用户数据稀疏、准确率低等问题。针对该问题,提出一种基于用户隐式行为的最大熵推荐算法。收集用户的历史操作信息,分别从用户、商品、用户-商品3个角度提取用户隐式行为特征;考虑到特征的有效性,利用Tree Ensemble Models对特征进行筛选和组合,构建行为模型挖掘用户潜在兴趣,完善用户缺失信息;针对特征之间的相关性问题,以最大熵原理构建特征函数,对用户进行商品推荐。在阿里移动推荐算法数据集上的仿真结果表明,所提算法可以有效解决数据稀疏性问题,提高推荐准确率。
In the field of E-commerce recommendation,since users always have low frequency of purchase behavior and seldom leave the evaluation information,the recommendation system faces the problem of sparseness of user data and the low accuracy of recommendation.To solve the problem,a maximum entropy recommendation algorithm based on user implicit behavior features was proposed.The user's historical operation information was collected and the user behavior features were extracted from the directions of user,item as well as user-item.With the help of Tree Ensemble Models,the features were selected and combined for better effectiveness and the potential interest of users was mined by applying the behavior model to complete the lost information of users.For the issue of correlation between features,the maximum entropy model was used to build feature functions to recommend for users.The simulation on dataset of Ali mobile recommendation algorithm shows that the proposed algorithm can effectively solve the problem of data sparsity and improve the accuracy of recommendation.
In the field of E-commerce recommendation,since users always have low frequency of purchase behavior and seldom leave the evaluation information,the recommendation system faces the problem of sparseness of user data and the low accuracy of recommendation.To solve the problem,a maximum entropy recommendation algorithm based on user implicit behavior features was proposed.The user's historical operation information was collected and the user behavior features were extracted from the directions of user,item as well as user-item.With the help of Tree Ensemble Models,the features were selected and combined for better effectiveness and the potential interest of users was mined by applying the behavior model to complete the lost information of users.For the issue of correlation between features,the maximum entropy model was used to build feature functions to recommend for users.The simulation on dataset of Ali mobile recommendation algorithm shows that the proposed algorithm can effectively solve the problem of data sparsity and improve the accuracy of recommendation.
引文
[1]HUANG Hongcheng,SUN Xinran,HU Min.An information diffusion model of social network based on node attitude[J].Advanced Engineering Sciences,2018,50(1):113-119(in Chinese).[黄宏程,孙欣然,胡敏.基于节点态度的社交网络信息传播模型[J].工程科学与技术,2018,50(1):113-119.]
[2]LI Penghui.Research on e-commerce personalized recommendation system based on mobile platform[D].Dalian:Dalian Maritime University,2014(in Chinese).[李鹏辉.基于移动平台的电商个性化推荐系统研究[D].大连:大连海事大学,2014.]
[3]Liu H,Hu Z,Mian A,et al.A new user similarity model to improve the accuracy of collaborative filtering[J].KnowledgeBased Systems,2014,56(3):156-166.
[4]Bessa A,Santos RLT,Veloso A,et al.Exploiting item coutility to improve collaborative filtering recommendations[J].Journal of the Association for Information Science and Technology,2017,68(10):2380-2393.
[5]Chen K,Zhang Z,Long J,et al.Turning from TF-IDF to TF-IGM for term weighting in text classification[J].Expert Systems with Applications an International Journal,2016,66(C):245-260.
[6]YANG Wu,TANG Rui,LU Ling.News recommendation method by fusion of content-based recommendation and collaborative filtering[J].Journal of Computer Applications,2016,36(2):414-418(in Chinese).[杨武,唐瑞,卢玲.基于内容的推荐与协同过滤融合的新闻推荐方法[J].计算机应用,2016,36(2):414-418.]
[7]LIU Shuai, YANG Yingjie,CHANG Dexian,et al.Improved fuzzy association rule and its mining algorithm[J].Computer Engineering and Design,2015,36(4):942-946(in Chinese).[刘帅,杨英杰,常德显,等.改进的模糊关联规则及其挖掘算法[J].计算机工程与设计,2015,36(4):942-946.]
[8]HU Weihua,FENG Wei.Improved apriori algorithm based on decomposed transaction matrix[J].Journal of Computer Applications,2014,34(S2):113-116(in Chinese).[胡维华,冯伟.基于分解事务矩阵的关联规则挖掘算法[J].计算机应用,2014,34(S2):113-116.]
[9]Zhang HR,Min F.Three-way recommender systems based on random forests[J]. Knowledge-Based Systems,2016,91(C):275-286.
[10]Guo L,Ma J,Jiang HR,et al.Social trust aware item recommendation for implicit feedback[J].Journal of Computer Science and Technology,2015,30(5):1039-1053.
[11]HE Jieyue,MA Bei.Based on real-valued conditional restricted Boltzmann machine and socail network for collaborative filtering[J].Chinese Journal of Computers,2016,39(1):183-195(in Chinese).[何洁月,马贝.利用社交关系的实值条件受限玻尔兹曼机协同过滤推荐算法[J].计算机学报,2016,39(1):183-195.]
[12]MA Jianwei,CHEN Honghui,STEPHAN Reiff-Marganiec.Recommending services via hybrid recommendation algorithms and hidden Markov model in cloud[J].Journal of Central South University(Science and Technology),2016,47(1):82-90(in Chinese).[马建威,陈洪辉,STEPHAN ReiffMarganiec.基于混合推荐和隐马尔科夫模型的服务推荐方法[J].中南大学学报(自然科学版),2016,47(1):82-90.]
[13]Li D,Zhao G,Wang Z,et al.A method of purchase prediction based on user behavior log[C]//IEEE International Conference on Data Mining Workshop.New York:IEEE,2015:1031-1039.
[14]WANG Yuheng.Improvement and application of random forest algorithm in recomender systems[D].Hangzhou:Zhejiang University,2016:15-23(in Chinese).[王宇恒.推荐系统中随机森林算法的优化与应用[D].杭州:浙江大学,2016:15-23.]
[15]MAO Jiumei.Research on collaborative filtering recommendation based on heterogeneous information network[D].Suzhou:Soochow University,2014:1(in Chinese).[冒九妹.基于异构信息网络的协同过滤推荐技术研究[D].苏州:苏州大学,2014:1.]
[16]XU Hong.Research of link prediction in social network based on probability model[D].Guangzhou:South China University of Technology,2016:28(in Chinese).[徐宏.基于概率模型的社会网络链接预测研究[D].广州:华南理工大学,2016:28.]
[2]LI Penghui.Research on e-commerce personalized recommendation system based on mobile platform[D].Dalian:Dalian Maritime University,2014(in Chinese).[李鹏辉.基于移动平台的电商个性化推荐系统研究[D].大连:大连海事大学,2014.]
[3]Liu H,Hu Z,Mian A,et al.A new user similarity model to improve the accuracy of collaborative filtering[J].KnowledgeBased Systems,2014,56(3):156-166.
[4]Bessa A,Santos RLT,Veloso A,et al.Exploiting item coutility to improve collaborative filtering recommendations[J].Journal of the Association for Information Science and Technology,2017,68(10):2380-2393.
[5]Chen K,Zhang Z,Long J,et al.Turning from TF-IDF to TF-IGM for term weighting in text classification[J].Expert Systems with Applications an International Journal,2016,66(C):245-260.
[6]YANG Wu,TANG Rui,LU Ling.News recommendation method by fusion of content-based recommendation and collaborative filtering[J].Journal of Computer Applications,2016,36(2):414-418(in Chinese).[杨武,唐瑞,卢玲.基于内容的推荐与协同过滤融合的新闻推荐方法[J].计算机应用,2016,36(2):414-418.]
[7]LIU Shuai, YANG Yingjie,CHANG Dexian,et al.Improved fuzzy association rule and its mining algorithm[J].Computer Engineering and Design,2015,36(4):942-946(in Chinese).[刘帅,杨英杰,常德显,等.改进的模糊关联规则及其挖掘算法[J].计算机工程与设计,2015,36(4):942-946.]
[8]HU Weihua,FENG Wei.Improved apriori algorithm based on decomposed transaction matrix[J].Journal of Computer Applications,2014,34(S2):113-116(in Chinese).[胡维华,冯伟.基于分解事务矩阵的关联规则挖掘算法[J].计算机应用,2014,34(S2):113-116.]
[9]Zhang HR,Min F.Three-way recommender systems based on random forests[J]. Knowledge-Based Systems,2016,91(C):275-286.
[10]Guo L,Ma J,Jiang HR,et al.Social trust aware item recommendation for implicit feedback[J].Journal of Computer Science and Technology,2015,30(5):1039-1053.
[11]HE Jieyue,MA Bei.Based on real-valued conditional restricted Boltzmann machine and socail network for collaborative filtering[J].Chinese Journal of Computers,2016,39(1):183-195(in Chinese).[何洁月,马贝.利用社交关系的实值条件受限玻尔兹曼机协同过滤推荐算法[J].计算机学报,2016,39(1):183-195.]
[12]MA Jianwei,CHEN Honghui,STEPHAN Reiff-Marganiec.Recommending services via hybrid recommendation algorithms and hidden Markov model in cloud[J].Journal of Central South University(Science and Technology),2016,47(1):82-90(in Chinese).[马建威,陈洪辉,STEPHAN ReiffMarganiec.基于混合推荐和隐马尔科夫模型的服务推荐方法[J].中南大学学报(自然科学版),2016,47(1):82-90.]
[13]Li D,Zhao G,Wang Z,et al.A method of purchase prediction based on user behavior log[C]//IEEE International Conference on Data Mining Workshop.New York:IEEE,2015:1031-1039.
[14]WANG Yuheng.Improvement and application of random forest algorithm in recomender systems[D].Hangzhou:Zhejiang University,2016:15-23(in Chinese).[王宇恒.推荐系统中随机森林算法的优化与应用[D].杭州:浙江大学,2016:15-23.]
[15]MAO Jiumei.Research on collaborative filtering recommendation based on heterogeneous information network[D].Suzhou:Soochow University,2014:1(in Chinese).[冒九妹.基于异构信息网络的协同过滤推荐技术研究[D].苏州:苏州大学,2014:1.]
[16]XU Hong.Research of link prediction in social network based on probability model[D].Guangzhou:South China University of Technology,2016:28(in Chinese).[徐宏.基于概率模型的社会网络链接预测研究[D].广州:华南理工大学,2016:28.]