感知用户年龄的Item-based协同过滤推荐算法
|
摘要
随着大数据时代的到来,推荐系统为人们寻找自己感兴趣的物品或事件提供了捷径。协同过滤推荐算法分为User-based协同过滤算法和Item-based协同过滤推荐算法。传统Item-based协同过滤推荐算法只关注Item间的相似度,与目标用户特征无关,因此传统算法相似度不能有效反映Item间的相似程度,推荐准确率低。并且传统Item-based协同过滤算法需要基于所有用户的历史行为数据进行计算,随着数据量的快速增长计算量不断增大,推荐时效性差。针对以上问题,提出了一种感知用户年龄的Item-based协同过滤推荐算法,基于用户年龄特征对用户进行分类,在类内采用加权相似度对Item间的相似度进行计算,并且在Spark分布式计算平台上运行测试。实验结果显示,该算法不仅保证了推荐准确率,而且大幅度提高了推荐效率,提升了推荐系统的实时性。
With the advent of the big data era,the recommendation system provides a shortcut for people to find objects or events that they are interested in. Collaborative filtering recommendation algorithms are divided into user-based collaborative filtering algorithm and item-based collaborative filtering recommendation algorithm. The traditional Item-based collaborative filtering recommendation algorithm only pays attention to the similarity between the items and has nothing to do with the characteristics of the target user. And the similarity of traditional algorithms cannot effectively reflect the degree of similarity between items,which leads to inaccurate recommendations. The traditional Item-based collaborative filtering algorithm needs to be calculated based on all users' historical behavior data. With the rapid increase of the amount of data,the amount of calculation continues to increase and the recommendation timeliness is poor. For this,we propose a user's age-aware item-based collaborative filtering recommendation algorithm. Users are classified based on user age characteristics. The similarity between items is calculated by weighted similarity within the class and the running test is implemented on the Spark distributed computing platform. Experiment shows that the proposed algorithm can greatly improve the recommendation efficiency and the real-time performance of the recommendation system while ensuring the accuracy of the recommendation.
With the advent of the big data era,the recommendation system provides a shortcut for people to find objects or events that they are interested in. Collaborative filtering recommendation algorithms are divided into user-based collaborative filtering algorithm and item-based collaborative filtering recommendation algorithm. The traditional Item-based collaborative filtering recommendation algorithm only pays attention to the similarity between the items and has nothing to do with the characteristics of the target user. And the similarity of traditional algorithms cannot effectively reflect the degree of similarity between items,which leads to inaccurate recommendations. The traditional Item-based collaborative filtering algorithm needs to be calculated based on all users' historical behavior data. With the rapid increase of the amount of data,the amount of calculation continues to increase and the recommendation timeliness is poor. For this,we propose a user's age-aware item-based collaborative filtering recommendation algorithm. Users are classified based on user age characteristics. The similarity between items is calculated by weighted similarity within the class and the running test is implemented on the Spark distributed computing platform. Experiment shows that the proposed algorithm can greatly improve the recommendation efficiency and the real-time performance of the recommendation system while ensuring the accuracy of the recommendation.
引文
[1] NAIMI A I,WESTREICH D J.Big data:a revolution that will transform how we live,work,and think[J].American Journal of Epidemiology,2014,179(9):1143-1144.
[2] BA Qilong,LI Xiaoyong,BAI Zhongying.Clustering colla- borative filtering recommendation system based on SVD algorithm[C]//IEEE international conference on software engineering and service science.Beijing,China:IEEE,2013:963-967.
[3] 高发展,黄梦醒,张婷婷.综合用户特征及专家信任的协作过滤推荐算法[J].计算机科学,2017,44(2):103-106.
[4] 谢霖铨,梁博群.结合用户特征分类和动态时间的协同过滤推荐[J].计算机工程与应用,2017,53(6):80-84.
[5] 吴湖,王永吉,王哲,等.两阶段联合聚类协同过滤算法[J].软件学报,2010,21(5):1042-1054.
[6] ZHAO Zhidan,SHANG Mingsheng.User-based collaborative-filtering recommendation algorithms on Hadoop[C]//International conference on knowledge discovery and data mining.Phuket,Thailand:IEEE,2010:478-481.
[7] 田保军,胡培培,杜晓娟,等.Hadoop下基于聚类协同过滤推荐算法优化的研究[J].计算机工程与科学,2016,38(8):1615-1624.
[8] 罗辛,欧阳元新,熊璋,等.通过相似度支持度优化基于K近邻的协同过滤算法[J].计算机学报,2010,33(8):1437-1445.
[9] 明小红.基于用户聚类的协同过滤推荐算法研究[D].北京:北京交通大学,2017.
[10] 徐江辉.基于Hadoop的聚类协同过滤推荐算法研究及应用[D].长沙:湖南大学,2016.
[11] 唐积益.推荐系统中相似度计算方法的研究[D].镇江:江苏科技大学,2015.
[12] 廖彬,张陶,国冰磊,等.基于Spark的ItemBased推荐算法性能优化[J].计算机应用,2017,37(7):1900-1905.
[13] JIN Chen,LIU Ruoqian,CHEN Zhengzhang,et al.A scalable hierarchical clustering algorithm using spark[C]//IEEE first international conference on big data computing service and applications.Redwood City,CA,USA:IEEE,2015:418-426.
[14] KARAU H.Spark快速大数据分析[M].北京:人民邮电出版社,2015.
[15] KURDIJA A S,SILIC M,VLADIMIR K,et al.Efficient global correlation measures for a collaborative filtering dataset[J].Knowledge-Based Systems,2018,147(1):36-42.
[2] BA Qilong,LI Xiaoyong,BAI Zhongying.Clustering colla- borative filtering recommendation system based on SVD algorithm[C]//IEEE international conference on software engineering and service science.Beijing,China:IEEE,2013:963-967.
[3] 高发展,黄梦醒,张婷婷.综合用户特征及专家信任的协作过滤推荐算法[J].计算机科学,2017,44(2):103-106.
[4] 谢霖铨,梁博群.结合用户特征分类和动态时间的协同过滤推荐[J].计算机工程与应用,2017,53(6):80-84.
[5] 吴湖,王永吉,王哲,等.两阶段联合聚类协同过滤算法[J].软件学报,2010,21(5):1042-1054.
[6] ZHAO Zhidan,SHANG Mingsheng.User-based collaborative-filtering recommendation algorithms on Hadoop[C]//International conference on knowledge discovery and data mining.Phuket,Thailand:IEEE,2010:478-481.
[7] 田保军,胡培培,杜晓娟,等.Hadoop下基于聚类协同过滤推荐算法优化的研究[J].计算机工程与科学,2016,38(8):1615-1624.
[8] 罗辛,欧阳元新,熊璋,等.通过相似度支持度优化基于K近邻的协同过滤算法[J].计算机学报,2010,33(8):1437-1445.
[9] 明小红.基于用户聚类的协同过滤推荐算法研究[D].北京:北京交通大学,2017.
[10] 徐江辉.基于Hadoop的聚类协同过滤推荐算法研究及应用[D].长沙:湖南大学,2016.
[11] 唐积益.推荐系统中相似度计算方法的研究[D].镇江:江苏科技大学,2015.
[12] 廖彬,张陶,国冰磊,等.基于Spark的ItemBased推荐算法性能优化[J].计算机应用,2017,37(7):1900-1905.
[13] JIN Chen,LIU Ruoqian,CHEN Zhengzhang,et al.A scalable hierarchical clustering algorithm using spark[C]//IEEE first international conference on big data computing service and applications.Redwood City,CA,USA:IEEE,2015:418-426.
[14] KARAU H.Spark快速大数据分析[M].北京:人民邮电出版社,2015.
[15] KURDIJA A S,SILIC M,VLADIMIR K,et al.Efficient global correlation measures for a collaborative filtering dataset[J].Knowledge-Based Systems,2018,147(1):36-42.