融合时间和类型特征加权的矩阵分解推荐算法
|
摘要
针对推荐算法的信息过期问题,结合遗忘函数和信息保持期的改进时间权重引入矩阵分解模型,提出一种基于改进时间权重的矩阵分解协同过滤算法(MFTWCF,MF-based and improved time weighted collabora tive filtering),相比前人提出的基于改进时间权重的邻域协同过滤算法(NTWCF,neighborhood-based and improved time weighted collaboratire filering algorithm),准确性显著提升了26.58%。由于过去的信息所包含的特征在随后的时间里可能被用户持续关注,从而增强过期信息对推荐的影响力,所以提出了融合时间权重和类型影响力加强权重的改进算法(MFTTWCF,MF-bosed and imporved time and type weighteel collaborative filtering)修正上述时间权重。电影数据集的实验证明,MFTTWCF算法预测的准确性比MFTWCF算法提高了3.58%,能够取得更好的推荐效果,适用于通过预测评分进行推荐的系统。
In order to solve the problem of information expiration of the recommender systems,we introduced the improved time weight of forgetting function and information retention period into matrix factorization model(MF)and proposed a MF-based and improved-time weighted collaborative filtering algorithm(MFTWCF)whose prediction accuracy had been raised by about 26.58% compared with that of neighborhood-based and improved-time weighted collaborative filtering algorithm(NTWCF).In view of the facts that users could continuously get access to some characteristics of past information,which would have greater influence for recommendation,we proposed the type weight to strengthen the information influence and to correct the improved time weight in MFTWCF.The new improved algorithm is called MF-based improved-time and type weighted collaborative filtering algorithm(MFTTWCF).The results of movie dataset experiments show that the prediction accuracy of MFTTWCF algorithm is 3.58% higher than that of MFTWCF algorithm and can achieve better recommendation effect.And it is applicable to recommender systems with rating prediction.
In order to solve the problem of information expiration of the recommender systems,we introduced the improved time weight of forgetting function and information retention period into matrix factorization model(MF)and proposed a MF-based and improved-time weighted collaborative filtering algorithm(MFTWCF)whose prediction accuracy had been raised by about 26.58% compared with that of neighborhood-based and improved-time weighted collaborative filtering algorithm(NTWCF).In view of the facts that users could continuously get access to some characteristics of past information,which would have greater influence for recommendation,we proposed the type weight to strengthen the information influence and to correct the improved time weight in MFTWCF.The new improved algorithm is called MF-based improved-time and type weighted collaborative filtering algorithm(MFTTWCF).The results of movie dataset experiments show that the prediction accuracy of MFTTWCF algorithm is 3.58% higher than that of MFTWCF algorithm and can achieve better recommendation effect.And it is applicable to recommender systems with rating prediction.
引文
[1]Katarya R,Verma O P.Recent developments in affective recommender systems[J].Physica A:Statistical Mechanics and its Applications,2016,461:182-190.
[2]Lu J,Wu D S,Mao M S,et al.Recommender system application developments:a survey[J].Decision Support Systems,2015,74(C):12-32.
[3]Chen L,Chen G L,Wang F.Recommender systems based on user reviews:the state of the art[J].User Modeling and User-Adapted Interaction,2015,25(2):99-154.
[4]Ricci F,Rokach L,Shapira B.Recommender systems handbook[M].US:Springer,2015.
[5]Wang B L,Huang J H,Ou L B,et al.A collaborative filtering algorithm fusing user-based,item-based and social networks[C]∥IEEE International Conference on Big Data,IEEE Big Data 2015.Piscataway,NJ,USA:IEEE,2015:2337-2343.
[6]Ma X,Wang C,Yu Q,et al.An FPGA-based accelerator for neighborhood-based collaborative filtering recommendation algorithms[C]∥IEEE International Conference on Cluster Computing,CLUSTER 2015.Piscataway,NJ,USA:IEEE,2015:494-495.
[7]郑翠翠,李林.协同过滤算法中的相似性度量方法研究[J].计算机工程与应用,2014,50(8):147-149.ZHENG Cuicui,LI Lin.Research on method of similarity measurement in collaborative filter algorithm[J].Computer Engineering and Applications,2014,50(8):147-149.(in Chinese)
[8]Aggarwal C C.Model-Based Collaborative Filtering[C]∥Recommender Systems.Cham:Springer International Publishing,2016:71-138.
[9]孟祥武,刘树栋,张玉洁,等.社会化推荐系统研究[J].软件学报,2015,26(6):1356-1372.MENG Xiangwu,LIU Shudong,ZHANG Yujie,et al.Research on social recommender systems[J].Journal of Software,2015,26(6):1356-1372.(in Chinese)
[10]Bokde D,Girase S,Mukhopadhyay D.Matrix factorization model in collaborative filtering algorithms:a survey[J].Procedia Computer Science,2015,49:136-146.
[11]Guo H,Tang R,Ye Y M,et al.Deepfm:A factorization-machine based neural network for CTR prediction[J].International Joint Conferences on Artificial Intelligence,2017:1725-1731.
[12]Barry Crabtree I,Soltysiak S J.Identifying and tracking changing interests[J].International Journal on Digital Libraries,1998,2(1):38-53.
[13]Koychev I,Schwab I.Adaptation to drifting user's interests[J].Proceedings of Ecml Workshop Machine Learning in New Information Age,2000:39-46.
[14]Koychev I.Gradual forgetting for adaptation to concept drift[C]∥Proceedings of ECAI 2000Workshop Current Issues in Spatio-Temporal Reasoning,Berlin:IEEE,2000:101-107.
[15]Ding Y,Li X.Time weight collaborative filtering[C]∥CIKM’05-Proceedings of the 14th ACM International Conference on Information and Knowledge Management.New York,USA:ACM,2005,1:485-492.
[16]邢春晓,高凤荣,战思南,等.适应用户兴趣变化的协同过滤推荐算法[J].计算机研究与发展,2007,44(2):296-301.XING Chunxiao,GAO Fengrong,ZHAN Sinan,et al.A collaborative filtering recommendation algorithm incorporated with user interest change[J].Journal of Computer Research and Development,2007,44(2):296-301.(in Chinese)
[17]Rendle S.Social network and click-through prediction with factorization machines[C]∥KDD-Cup Workshop.Beijing,China:IEEE,2012.
[18]赵海燕,王颖,陈庆奎,等.产品时效性感知的个性化推荐算法[J].小型微型计算机系统,2017,38(9):2022-2027.ZHAO Haiyan,WANG Ying,CHEN Qingkui,et al.Timeliness sensitive collective matrix factorization personalized recommendation[J].Journal of Chinese Computer Systems,2017,38(9):2022-2027.(in Chinese)
[19]Sun B S,Dong L Y.Dynamic model adaptive to user interest drift based on cluster and nearest neighbors[J].IEEEAccess,2017,5:1682-1691.
[20]兰艳,曹芳芳.面向电影推荐的时间加权协同过滤算法的研究[J].计算机科学,2017,44(4):295-301.LAN Yan,CAO Fangfang.Research of time weighted collaborative filtering algorithm in movie recommendation[J].Computer Science,2017,44(4):295-301.(in Chinese)
[21]Wei S X,Zheng X L,Chen D R,et al.A hybrid approach for movie recommendation via tags and ratings[J].Electronic Commerce Research and Applications,2016,18(C):83-94.
[2]Lu J,Wu D S,Mao M S,et al.Recommender system application developments:a survey[J].Decision Support Systems,2015,74(C):12-32.
[3]Chen L,Chen G L,Wang F.Recommender systems based on user reviews:the state of the art[J].User Modeling and User-Adapted Interaction,2015,25(2):99-154.
[4]Ricci F,Rokach L,Shapira B.Recommender systems handbook[M].US:Springer,2015.
[5]Wang B L,Huang J H,Ou L B,et al.A collaborative filtering algorithm fusing user-based,item-based and social networks[C]∥IEEE International Conference on Big Data,IEEE Big Data 2015.Piscataway,NJ,USA:IEEE,2015:2337-2343.
[6]Ma X,Wang C,Yu Q,et al.An FPGA-based accelerator for neighborhood-based collaborative filtering recommendation algorithms[C]∥IEEE International Conference on Cluster Computing,CLUSTER 2015.Piscataway,NJ,USA:IEEE,2015:494-495.
[7]郑翠翠,李林.协同过滤算法中的相似性度量方法研究[J].计算机工程与应用,2014,50(8):147-149.ZHENG Cuicui,LI Lin.Research on method of similarity measurement in collaborative filter algorithm[J].Computer Engineering and Applications,2014,50(8):147-149.(in Chinese)
[8]Aggarwal C C.Model-Based Collaborative Filtering[C]∥Recommender Systems.Cham:Springer International Publishing,2016:71-138.
[9]孟祥武,刘树栋,张玉洁,等.社会化推荐系统研究[J].软件学报,2015,26(6):1356-1372.MENG Xiangwu,LIU Shudong,ZHANG Yujie,et al.Research on social recommender systems[J].Journal of Software,2015,26(6):1356-1372.(in Chinese)
[10]Bokde D,Girase S,Mukhopadhyay D.Matrix factorization model in collaborative filtering algorithms:a survey[J].Procedia Computer Science,2015,49:136-146.
[11]Guo H,Tang R,Ye Y M,et al.Deepfm:A factorization-machine based neural network for CTR prediction[J].International Joint Conferences on Artificial Intelligence,2017:1725-1731.
[12]Barry Crabtree I,Soltysiak S J.Identifying and tracking changing interests[J].International Journal on Digital Libraries,1998,2(1):38-53.
[13]Koychev I,Schwab I.Adaptation to drifting user's interests[J].Proceedings of Ecml Workshop Machine Learning in New Information Age,2000:39-46.
[14]Koychev I.Gradual forgetting for adaptation to concept drift[C]∥Proceedings of ECAI 2000Workshop Current Issues in Spatio-Temporal Reasoning,Berlin:IEEE,2000:101-107.
[15]Ding Y,Li X.Time weight collaborative filtering[C]∥CIKM’05-Proceedings of the 14th ACM International Conference on Information and Knowledge Management.New York,USA:ACM,2005,1:485-492.
[16]邢春晓,高凤荣,战思南,等.适应用户兴趣变化的协同过滤推荐算法[J].计算机研究与发展,2007,44(2):296-301.XING Chunxiao,GAO Fengrong,ZHAN Sinan,et al.A collaborative filtering recommendation algorithm incorporated with user interest change[J].Journal of Computer Research and Development,2007,44(2):296-301.(in Chinese)
[17]Rendle S.Social network and click-through prediction with factorization machines[C]∥KDD-Cup Workshop.Beijing,China:IEEE,2012.
[18]赵海燕,王颖,陈庆奎,等.产品时效性感知的个性化推荐算法[J].小型微型计算机系统,2017,38(9):2022-2027.ZHAO Haiyan,WANG Ying,CHEN Qingkui,et al.Timeliness sensitive collective matrix factorization personalized recommendation[J].Journal of Chinese Computer Systems,2017,38(9):2022-2027.(in Chinese)
[19]Sun B S,Dong L Y.Dynamic model adaptive to user interest drift based on cluster and nearest neighbors[J].IEEEAccess,2017,5:1682-1691.
[20]兰艳,曹芳芳.面向电影推荐的时间加权协同过滤算法的研究[J].计算机科学,2017,44(4):295-301.LAN Yan,CAO Fangfang.Research of time weighted collaborative filtering algorithm in movie recommendation[J].Computer Science,2017,44(4):295-301.(in Chinese)
[21]Wei S X,Zheng X L,Chen D R,et al.A hybrid approach for movie recommendation via tags and ratings[J].Electronic Commerce Research and Applications,2016,18(C):83-94.