一种负样本改进的LDA主题模型推荐算法
|
摘要
LDA主题模型是文本挖掘领域的重要算法,同时在推荐系统当中也有不错的表现.通过LDA主题模型挖掘用户感兴趣的主题,是目前最常用的用户兴趣主题挖掘方法之一.为了提高LDA主题模型应用在推荐系统时的推荐质量,我们提出了一种基于负样本进行学习的方法 negLDA.通过创造出负样本来学习用户对物品的负面预测评分,同时结合正样本学习得到的正面预测评分,从正反两个方面进行综合评测,从而更加精确地衡量出用户对物品的预测评分.通过在MoviesLens-100k、MovieLens-1M、FilmTrust这三个数据集上的实验,表明所提出的算法在精确率、召回率、AUC三个指标上相比传统算法均有一定改进.
The LDA theme model is an important algorithm in the field of text mining,and it also has a good performance in the field of recommender systems.The LDA theme model is a popular topic mining method of mining user interested topics.In order to improve the recommended quality of the LDA theme model,we proposed a negLDA method based on negative samples.By creating negative samples to learn the user′s negative predictions to the items and combining with the positive samples to learn the positive predictive scores,we could more accurately measure the user's forecast scores for the items.Experiments on the three real-world datasets show that the algorithm proposed has been improved compared with the traditional algorithm in terms of the accuracy,the recall rate and AUC.
The LDA theme model is an important algorithm in the field of text mining,and it also has a good performance in the field of recommender systems.The LDA theme model is a popular topic mining method of mining user interested topics.In order to improve the recommended quality of the LDA theme model,we proposed a negLDA method based on negative samples.By creating negative samples to learn the user′s negative predictions to the items and combining with the positive samples to learn the positive predictive scores,we could more accurately measure the user's forecast scores for the items.Experiments on the three real-world datasets show that the algorithm proposed has been improved compared with the traditional algorithm in terms of the accuracy,the recall rate and AUC.
引文
[1]PAZZANI M J,BILLSUS D.Content-based recommendation systems[C]//Proc of the Adaptive Web.Heidelberg,Berlin:Springer-Verlag,2007:325-341.
[2]BALTRUNAS L,RICCI F.Experimental evaluation of context-dependent collaborative filtering using item splitting[J].User Modeling and User-Adapted Interaction,2014,24(1,2):7-34.
[3]CHEN W,NIU Z,ZHAO X,LI Y.A hybrid recommendation algorithm adapted in e-learning environments[J].World Wide Web,2014,17(2):271-284.
[4]SARWAR B M,KARYPIS G,KONSTAN J A,et al.A-nalysis of recommendation algorithms for ecommerce[C]//Procedings of the 2nd ACM Conference on Electronic Commerce.New York:ACM Press,2000:158-167.
[5]SARWAR B M,KARYPIS G,KONSTAN J,et al.Itembased collaborative filtering recommendation algorithms[C]//Proceding of the 10th International Conference on World Wide Web.New York:ACM Press,2001:285-295.
[6]KOREN Y,BEL R,VOLINSKY C.Matrix factorization techniques for recommender systems[J].Computer,2009,42(8):30-37
[7]黄璐,林川杰,何军,等.融合主题模型和协同过滤的多样化移动应用推荐[J].软件学报,2017,28(3):708-720.HUANG L,LIN C J,HE J,et al.Diversifled mobile app recommendation combining topic model and collaborative filtering[J].Journal of Software,2017,28(3):708-720.
[8]高娜,杨明.嵌入LDA主题模型的协同过滤推荐算法[J].计算机科学,2016,43(3):57-61.GAO N,YANG M.Topic model embedded in collaborative filtering recommendation algorithm[J].Computer Science,2016,43(3):57-61.
[9]彭敏,席俊杰,代心媛,等.基于情感分析和LDA主题模型的协同过滤推荐算法[J].中文信息学报,2017,31(2):194-203.PENG M,XI J J,DAI X Y.et al.Collaborative filtering recommendation based on sentiment analysis and LDA topic model[J].Journal of Chinese information processing,2017,31(2):194-203.
[10]BLEI D M,NG A Y,JORDAN M I.Latent dirichlet alocation[J].Journal of machine learning research,2003,3:993-1022.
[11]HEINRICH G.Parameter estimation for text analysis[R].Germany:University of Leipzig,2008.
[12]HOFMANN T.Latent semantic models for collaborative filtering[J].ACM Transactions on Information Systems,2004,22(1):89-115.
[13]RENDLE S,FREUDENTHALER C,GANTNER Z,et al.BPR:Bayesian personalized ranking from implicit feedback[C]//Proceedings of the 25th Conference on Uncertainty in Artificial Intelligence(UAI 09).Arlington,Virginia,USA:AUAI Press,2009:452-461.
[2]BALTRUNAS L,RICCI F.Experimental evaluation of context-dependent collaborative filtering using item splitting[J].User Modeling and User-Adapted Interaction,2014,24(1,2):7-34.
[3]CHEN W,NIU Z,ZHAO X,LI Y.A hybrid recommendation algorithm adapted in e-learning environments[J].World Wide Web,2014,17(2):271-284.
[4]SARWAR B M,KARYPIS G,KONSTAN J A,et al.A-nalysis of recommendation algorithms for ecommerce[C]//Procedings of the 2nd ACM Conference on Electronic Commerce.New York:ACM Press,2000:158-167.
[5]SARWAR B M,KARYPIS G,KONSTAN J,et al.Itembased collaborative filtering recommendation algorithms[C]//Proceding of the 10th International Conference on World Wide Web.New York:ACM Press,2001:285-295.
[6]KOREN Y,BEL R,VOLINSKY C.Matrix factorization techniques for recommender systems[J].Computer,2009,42(8):30-37
[7]黄璐,林川杰,何军,等.融合主题模型和协同过滤的多样化移动应用推荐[J].软件学报,2017,28(3):708-720.HUANG L,LIN C J,HE J,et al.Diversifled mobile app recommendation combining topic model and collaborative filtering[J].Journal of Software,2017,28(3):708-720.
[8]高娜,杨明.嵌入LDA主题模型的协同过滤推荐算法[J].计算机科学,2016,43(3):57-61.GAO N,YANG M.Topic model embedded in collaborative filtering recommendation algorithm[J].Computer Science,2016,43(3):57-61.
[9]彭敏,席俊杰,代心媛,等.基于情感分析和LDA主题模型的协同过滤推荐算法[J].中文信息学报,2017,31(2):194-203.PENG M,XI J J,DAI X Y.et al.Collaborative filtering recommendation based on sentiment analysis and LDA topic model[J].Journal of Chinese information processing,2017,31(2):194-203.
[10]BLEI D M,NG A Y,JORDAN M I.Latent dirichlet alocation[J].Journal of machine learning research,2003,3:993-1022.
[11]HEINRICH G.Parameter estimation for text analysis[R].Germany:University of Leipzig,2008.
[12]HOFMANN T.Latent semantic models for collaborative filtering[J].ACM Transactions on Information Systems,2004,22(1):89-115.
[13]RENDLE S,FREUDENTHALER C,GANTNER Z,et al.BPR:Bayesian personalized ranking from implicit feedback[C]//Proceedings of the 25th Conference on Uncertainty in Artificial Intelligence(UAI 09).Arlington,Virginia,USA:AUAI Press,2009:452-461.