针对协同过滤推荐系统的混淆托攻击模型
|
摘要
为了躲避现有的托攻击检测,攻击者利用混淆技术降低攻击概貌和普通概貌之间的区别,使之成为更多用户的近邻进而影响推荐系统的预测评分。对物品按属性划分,在混淆技术的基础上,给出噪音系数、评分偏移函数和目标偏移函数,提出混淆流行交叉托攻击模型。设计攻击概貌自动产生器注入系统数据库,同最流行项中添加平均项构造的混淆托攻击对比。实验结果表明,该托攻击模型对推荐系统的危害性更大,应加强防范。
In order to avoid the existing attack detection,the attacker uses the obfuscation techniques to reduce the difference between the attack profiles and the general profiles,and then to become more users neighbor,the predictive score of the recom-mended system is affected. Items are classified by attributes,the noise coefficient,the score offset function and the target offset function are given on the basis of confusing techniques. The popular crossing model of confusion is proposed. Attack profiles automatic generator is designed to inject system database,this model is compared with average-over-popular. The experimental result shows that the attack model is more harmful to the recommender system,prevention should be strengthened for recommender system.
In order to avoid the existing attack detection,the attacker uses the obfuscation techniques to reduce the difference between the attack profiles and the general profiles,and then to become more users neighbor,the predictive score of the recom-mended system is affected. Items are classified by attributes,the noise coefficient,the score offset function and the target offset function are given on the basis of confusing techniques. The popular crossing model of confusion is proposed. Attack profiles automatic generator is designed to inject system database,this model is compared with average-over-popular. The experimental result shows that the attack model is more harmful to the recommender system,prevention should be strengthened for recommender system.
引文
[1]石坤,余强,朱秋月.融合信息熵与兴趣度的协同过滤算法[J].计算机与数字工程,2017,45(2):338-342.SHI kun,YU Qiang,ZHU Qiuyue.A Collaborative Filter-ing Algorithm of Weighted Information Entropy and Ratingof User Interests[J].Computer&Digital Engineering,2017,45(2):338-342.
[2]朱扬勇,孙婧.推荐系统研究进展[J].计算机科学与探索,2015,9(5):513-525.ZHU Yangyong,SUN Jing.Recommender System:Up toNow[J].Journal of Frontiers of Computer Science andTechnology,2015,9(5):513-525.
[3]吕成戍.基于特征选择和支持向量机的托攻击检测方法[J].计算机应用与软件,2015,32(5):270-272.LV Chengshu.Shilling Attack Detection Based on FeatureSelection and SVM[J].Computer Applications and Soft-ware,2015,32(5):270-272.
[4]伍之昂,王有权,曹杰.推荐系统托攻击模型与检测技术[J].科学通报,2014,59(7):551-560.WU Zhiang,WANG Youquan,CAO Jie.A Survery on Shil-ling Attack Models and Detection Techniques for Recom-mender Systems[J].Chinese Science Bulletin,2014,59(7):551-560.
[5]黄光球,刘嘉飞.基于记忆原理的推荐系统托攻击检测模型[J].计算机工程,2012,38(5):25-29.HUANG Guangqiu,LIU Jiafei.Shilling Attack DetectionModel for Recommender System Based on Memory Princi-ple[J].Computer Engineering,2012,38(5):25-29.
[6]李聪.协同过滤推荐系统托攻击防御技术研究[D].长沙:国防科技大学,2012.LI Cong.Researches on Defense Technology against Shil-ling Attack in Collaborative Filtering Recommender Sys-tems[D].Changsha:National University of Defense Tech-nology,2012.
[7]李文涛,高旻,李华,等.一种基于流行度分类特征的托攻击检测算法[J].自动化学报,2015,41(9):1563-1576.LI Wentao,GAO Min,LI Hua,et al.An Shilling AttackDetection Algorithm Based on Popularity Degree Features.Acta Automatica Sinica,2015,41(9):1563-1576.
[8]Chirita P A,Nejdl W,Zamfir C.Preventing shilling at-tacks in online recommender systems[C]//ACM Interna-tional Workshop on Web Information and Data Manage-ment.New York:ACM,2005:67-74.
[9]伍之昂,庄毅,王有权,等.基于特征选择的推荐系统托攻击检测算法[J].电子学报,2012,40(8):1687-1693.WU Zhiang,ZHUANG Yi,WANG Youquan,et al.ShillingAttack Detection Based on Feature Selection for Recom-mender Systems.Acta Electronica Sinica,2012,40(8):1687-1693.
[10]彭飞,曾学文,邓浩江,等.基于特征子集的推荐系统托攻击无监督检测[J].计算机工程,2014,40(5):109-114.PENG Fei,ZENG Xuewen,DENG Haojiang,et al.Un-supervised Detection of Shilling Attack for Recommend-er System Based on Feature Subset.Computer Engineer-ing,2014(5):109-114.
[11]Cheng Z P,Zhang M.Statistical attack detection[C]//Proceedings of the 3rd International Conference on Rec-ommender Systems.New York:ACM.2009:149-156.
[12]Williams C,Mobasher B,Burke R,et al.Detection ofobfuscated attacks in collaborative recommender systems[C]//Proceedings of the ECAI 2006 Workshop on Recommender Systems.Ohmsha:IOS,2006:19-23.
[13]卫星君.针对协同过滤推荐系统的推举托攻击研究[D].西安:西安建筑科技大学,2016.WEI Xingjun.Researches on Push Shilling Attacks inCollaborative Filtering Recommender Systems[D].Xi'an:University of Architecture&technology,2016.
[14]Williams C A,Research Advisor,Mobasher B.Thesis:Profile Injection Attack Detection for Securing Collabora-tive Recommender Systems[J].Service Oriented Com-puting&Applications,2006,1(3):157-170.
[15]Kuzelewska U.Clustering Algorithms in Hybrid Recom-mender System on Movie Lens Data[J].Studies in LogicGrammar&Rhetoric,2014,37(1):125-139.
[16]万猛.推荐系统攻击检测算法的研究[D].成都:电子科技大学,2014.WAN Meng.The Research of Attack Detection Algo-rithm in Recommender System[D].Chengdu:Universityof Electronic Science and Technology of china,2014.
[2]朱扬勇,孙婧.推荐系统研究进展[J].计算机科学与探索,2015,9(5):513-525.ZHU Yangyong,SUN Jing.Recommender System:Up toNow[J].Journal of Frontiers of Computer Science andTechnology,2015,9(5):513-525.
[3]吕成戍.基于特征选择和支持向量机的托攻击检测方法[J].计算机应用与软件,2015,32(5):270-272.LV Chengshu.Shilling Attack Detection Based on FeatureSelection and SVM[J].Computer Applications and Soft-ware,2015,32(5):270-272.
[4]伍之昂,王有权,曹杰.推荐系统托攻击模型与检测技术[J].科学通报,2014,59(7):551-560.WU Zhiang,WANG Youquan,CAO Jie.A Survery on Shil-ling Attack Models and Detection Techniques for Recom-mender Systems[J].Chinese Science Bulletin,2014,59(7):551-560.
[5]黄光球,刘嘉飞.基于记忆原理的推荐系统托攻击检测模型[J].计算机工程,2012,38(5):25-29.HUANG Guangqiu,LIU Jiafei.Shilling Attack DetectionModel for Recommender System Based on Memory Princi-ple[J].Computer Engineering,2012,38(5):25-29.
[6]李聪.协同过滤推荐系统托攻击防御技术研究[D].长沙:国防科技大学,2012.LI Cong.Researches on Defense Technology against Shil-ling Attack in Collaborative Filtering Recommender Sys-tems[D].Changsha:National University of Defense Tech-nology,2012.
[7]李文涛,高旻,李华,等.一种基于流行度分类特征的托攻击检测算法[J].自动化学报,2015,41(9):1563-1576.LI Wentao,GAO Min,LI Hua,et al.An Shilling AttackDetection Algorithm Based on Popularity Degree Features.Acta Automatica Sinica,2015,41(9):1563-1576.
[8]Chirita P A,Nejdl W,Zamfir C.Preventing shilling at-tacks in online recommender systems[C]//ACM Interna-tional Workshop on Web Information and Data Manage-ment.New York:ACM,2005:67-74.
[9]伍之昂,庄毅,王有权,等.基于特征选择的推荐系统托攻击检测算法[J].电子学报,2012,40(8):1687-1693.WU Zhiang,ZHUANG Yi,WANG Youquan,et al.ShillingAttack Detection Based on Feature Selection for Recom-mender Systems.Acta Electronica Sinica,2012,40(8):1687-1693.
[10]彭飞,曾学文,邓浩江,等.基于特征子集的推荐系统托攻击无监督检测[J].计算机工程,2014,40(5):109-114.PENG Fei,ZENG Xuewen,DENG Haojiang,et al.Un-supervised Detection of Shilling Attack for Recommend-er System Based on Feature Subset.Computer Engineer-ing,2014(5):109-114.
[11]Cheng Z P,Zhang M.Statistical attack detection[C]//Proceedings of the 3rd International Conference on Rec-ommender Systems.New York:ACM.2009:149-156.
[12]Williams C,Mobasher B,Burke R,et al.Detection ofobfuscated attacks in collaborative recommender systems[C]//Proceedings of the ECAI 2006 Workshop on Recommender Systems.Ohmsha:IOS,2006:19-23.
[13]卫星君.针对协同过滤推荐系统的推举托攻击研究[D].西安:西安建筑科技大学,2016.WEI Xingjun.Researches on Push Shilling Attacks inCollaborative Filtering Recommender Systems[D].Xi'an:University of Architecture&technology,2016.
[14]Williams C A,Research Advisor,Mobasher B.Thesis:Profile Injection Attack Detection for Securing Collabora-tive Recommender Systems[J].Service Oriented Com-puting&Applications,2006,1(3):157-170.
[15]Kuzelewska U.Clustering Algorithms in Hybrid Recom-mender System on Movie Lens Data[J].Studies in LogicGrammar&Rhetoric,2014,37(1):125-139.
[16]万猛.推荐系统攻击检测算法的研究[D].成都:电子科技大学,2014.WAN Meng.The Research of Attack Detection Algo-rithm in Recommender System[D].Chengdu:Universityof Electronic Science and Technology of china,2014.