基于D-S证据理论的电子商务虚假评论者检测
|
摘要
在线商品的销售与商品评价信息密切相关,拥有较多好评信息的商品更受消费者的青睐.于是越来越多的电商商家开始雇佣甚至充当虚假评论者对商品进行不切实际的评论,广大消费者成为了最终的受害者.本文提出一种基于评论者行为的虚假评论者检测方法,该方法从虚假评论者作弊动机出发,综合考虑评论者评价行为、评论者交流行为以及评论者对商品的关注行为,将评论者行为视为证据并构建D-S证据理论模型.首先,本文利用多种维度对评论者的三种行为特征进行量化并构建三个独立的SVM模型,然后将SVM无阈值输出通过sigmoid函数实现后验概率输出,最后将其用于证据融合并根据识别框架下的证据支持度对评论者身份进行检测.实验结果表明,本文提出的方法准确有效.
Online product reviews can significantly affect quantity of sale,and consumers are more willing to buy products with high ratings and positive comments. Therefore,more and more online sellers began to hire or act as reviewspammers to write untruthful product reviews. There is no doubt that all of the consumers suffered from these fake reviews in the end. This paper proposes a method for reviewspammer detection based on reviewer's behaviors. We notice reviewspammer's cheating motivation and take reviewer's reviewbehavior,communicative behavior and their behavior of targeting specific commodities into consideration. Regarding these three behaviors as evidence,we propose a model based on the D-S evidence theory. Firstly,we define several dimensionalities to quantify reviewer's three types of behaviors and construct three different SVMmodels to reflect these behaviors. Secondly,we get posterior probability by inputting the unthresholded output of an SVMinto the sigmoid function. Then we create basic probability assignment by means of posterior probability and use the Dempster rules of evidence combination. Finally,we can detect reviewspammer on the basis of combined basic probability assignment. Experimental results showthat our method is effective and accurate in detecting reviewspammers.
Online product reviews can significantly affect quantity of sale,and consumers are more willing to buy products with high ratings and positive comments. Therefore,more and more online sellers began to hire or act as reviewspammers to write untruthful product reviews. There is no doubt that all of the consumers suffered from these fake reviews in the end. This paper proposes a method for reviewspammer detection based on reviewer's behaviors. We notice reviewspammer's cheating motivation and take reviewer's reviewbehavior,communicative behavior and their behavior of targeting specific commodities into consideration. Regarding these three behaviors as evidence,we propose a model based on the D-S evidence theory. Firstly,we define several dimensionalities to quantify reviewer's three types of behaviors and construct three different SVMmodels to reflect these behaviors. Secondly,we get posterior probability by inputting the unthresholded output of an SVMinto the sigmoid function. Then we create basic probability assignment by means of posterior probability and use the Dempster rules of evidence combination. Finally,we can detect reviewspammer on the basis of combined basic probability assignment. Experimental results showthat our method is effective and accurate in detecting reviewspammers.
引文
[1]Peng Lan,Zhou Qi-hai,Qiu Jiang-tao.Research on model of helpfulness factors of online customer reviews[J].Computer Science,2011,38(8):205-207.
[2]Mo Qian,Yang Ke.Overviewof web spammer detection[J].Journal of Software,2014,25(7):1505-1526.
[3]Lim Ee-Peng,Nguyen Viet-An,Jindal Nitin.Detecting product reviewspammers using rating behaviors[C].Proceedings of the 19th ACMInternational Conference on Information&Knowledge Management(CIKM),2010:939-948.
[4]Duan Xin-sheng.Evidence theory and its application on decision and artificial intelligence[M].Beijing:People's University of China Press,1993.
[5]Yang Jian-ping,Huang Hong-zhong,Miao Qiang,et al.A novel information fusion method based on Dempster-Shafer evidence theory for conflict resolution[J].Intelligent Data Analysis,2011,15(3):399-411.
[6]Yue Kun,Liu Wei-yi,Wang Xiao-ling,et al.An approach for measuring quality of web services based on superposition of uncertain factors[J].Journal of Computer Research and Development,2009,46(5):841-849.
[7]Zhuge Jian-wei,Wang Da-wei,Chen Yu,et al.A network anomaly detector based on the D-S evidence theory[J].Journal of Software,2006,17(3):463-471.
[8]Li Peng,Liu Si-feng.Interval-valued intuitionistic fuzzy numbers decision-making method based on grey incidence analysis and D-S theory of evidence[J].Acta Automatica Sinica,2011,37(8):993-998.
[9]Qiu Pei-yuan,Lu Feng,Zhang Heng-cai.Extracting traffic information from web texts with a D-S evidence theory based approach[C].Proceedings of the 21st International Conference on Geoinformatics,2013:1-5.
[10]Huang Shi-guo,Zhou Ming-quan,Geng Guo-hua.Angel invariant gabor-SVMfor insect recognition[J].Journal of Chinese Computer Systems,2010,31(1):143-146.
[11]Chen Kai-zhi,Le Cheng-pei,Zhong Shang-ping.Fusion method of distance metric learning and SVMfor image matching[J].Journal of Chinese Computer Systems,2015,36(6):1353-1357.
[12]Zhang Xiang,Xiao Xiao-ling,Xu Guang-you.Fuzzy support vector machine based on affinity among samples[J].Journal of Software,2006,17(5):951-958.
[13]Platt John C.Probabilistic outputs for support vector machines and comparisons to regularized likelihood methods[M].Alexander J.Smola.Advances in Large Margin Classifiers,Cambridge,MA:MIT Press,1999:1-11.
[14]Wang Guan,Xie Si-hong,Liu Bing,et al.Reviewgraph based online store reviewspammer[C].Proceedings of the 11th IEEE International Conference on Data Mining(ICDM),2011:1242-1247.
[15]Liang Dong-xu,Liu Xin-yue,Shen Hua.Detecting spam reviewers by combing reviewer feature and relationship[C].Proceedings of the International Conference on Informative&Cybernetics for Computational Social Systems(ICCSS),2014:102-107.
[16]Lu Yu-qing,Zhang Lei,Xiao Yu-dong,et al.Simultaneously detecting fake reviews and reviewspammers using factor graph model[C].Proceedings of the 5th ACMWeb Science Conference,2013:225-233.
[17]Rayana Shebuti,Akoglu Leman.Collective opinion spam detection bridging reviewnetworks and metadata[C].Proceedings of the21th ACMSIGKDD International Conference on Knowledge Discovery&Data Mining,Sydney,NSW,Australia,August 10-13,2015:985-994.
[18]Mukherjee Arjun,Kumar Abhinav,Liu Bing,et al.Spotting opinion spammers using behavioral footprints[C].Proceedings of the 19th ACMSIGKDD International Conference on Knowledge Discovery&Data Mining,2013:632-640.
[19]Xue Hao,Li Feng-jun,Seo Hyunjin,et al.Trust-Aware reviewspam detection[C].Proceedings of the 14th IEEE International Conference on Trust,Security and Privacy in Computing and Communications,2015,1:726-733.
[20]Ye Jun-ting,Kumar Santhosh,Akoglu Leman.Temporal opinion spam detection by multivariate indicative signals[C].Proceedings of the 10th International AAAI Conference on Web and Social Media,2016:743-746.
[21]Fei Ge-li,Mukherjee Arjun,Liu Bing,et al.Exploting burstiness in reviews for reviewspammer detection[C].Proceedings of the 7th International AAAI Conference on Weblogs and Social Media,2013:175-184.
[22]Xie Si-hong,Wang Guan,Lin Shu-yang,et al.Reviewspam detection via temporal pattern discovery[C].Proceedings of the 18th ACMSIGKDD International Conference on Knowledge Discovery&Data Mining,2012:823-831.
[23]Choo Euijin,Yu Ting,Chi Min.Detecting opinion spammer groups through community discovery and sentiment analysis[C].Proceedings of the IFIP Conference on Data&Applications Security&Privacy,2015,9149:170-187.
[24]Yang Min,Lu Zi-yu,Chen Xiao-jun,et al.Detecting reviewspammer groups[C].Proceedings of the 31st AAAI Conference on Artificial Intelligence,2017:5011-5012.
[25]Ye Jun-ting,Akoglu Leman.Discovering opinion spammer groups by network footprints[C].Proceedings of the Joint European Conference on Machine Learning&Knowledge Discovery in Databases,2015,9284:267-282.
[26]Salton Gerard.Experiments in automatic thesaurus construction for information retrieval[C].Proceedings of IFIP Congress,1971,71:115-123.
[27]Jacob Cohen.A coefficient of agreement for nominal scales[J].Educational&Psychological Measurement,1960,20(1):37-46.
[28]Kohavi Ron.A study of cross-validation and bootstrap for accuracy estimation and model selection[C].Proceedings of International Joint Conference on Artificial Intelligence(IJCAI),1995,14:1137-1143.
[29]Jarvelin Kalervo,Kekalainen Jaana.IR evaluation methods for retrieving highly relevant documents[C].Proceedings of the 29th international ACMSIGIR Conference on Research&Development in Information Retrieval,2000:41-48.
[1]彭岚,周启海,邱江涛.消费者在线评论有用性影响因素模型研究[J].计算机科学,2011,38(8):205-207.
[2]莫倩,杨珂.网络水军识别研究[J].软件学报,2014,25(7):1505-1526.
[4]段新生.证据理论与决策、人工智能[M].北京:中国人民大学出版社,1993.
[6]岳昆,刘惟一,王晓玲,等.一种基于不确定性因素叠加的W eb服务质量度量方法[J].计算机研究与发展,2009,46(5):841-849.
[7]诸葛建伟,王大为,陈昱,等.基于D-S证据理论的网络异常检测方法[J].软件学报,2006,17(3):463-471.
[8]李鹏,刘思峰.基于灰色关联分析和D-S证据理论的区间直觉模糊决策方法[J].自动化学报,2011,37(8):993-998.
[10]黄世国,周明全,耿国华.一种角度无关的Gabor-SVM昆虫识别[J].小型微型计算机系统,2010,31(1):143-146.
[11]陈开志,乐承沛,钟尚平.融合距离度量学习和SVM的图像匹配算法[J].小型微型计算机系统,2015,36(6):1353-1357.
[12]张翔,肖小玲,徐光祐.基于样本之间紧密度的模糊支持向量机方法[J].软件学报,2006,17(5):951-958.
[2]Mo Qian,Yang Ke.Overviewof web spammer detection[J].Journal of Software,2014,25(7):1505-1526.
[3]Lim Ee-Peng,Nguyen Viet-An,Jindal Nitin.Detecting product reviewspammers using rating behaviors[C].Proceedings of the 19th ACMInternational Conference on Information&Knowledge Management(CIKM),2010:939-948.
[4]Duan Xin-sheng.Evidence theory and its application on decision and artificial intelligence[M].Beijing:People's University of China Press,1993.
[5]Yang Jian-ping,Huang Hong-zhong,Miao Qiang,et al.A novel information fusion method based on Dempster-Shafer evidence theory for conflict resolution[J].Intelligent Data Analysis,2011,15(3):399-411.
[6]Yue Kun,Liu Wei-yi,Wang Xiao-ling,et al.An approach for measuring quality of web services based on superposition of uncertain factors[J].Journal of Computer Research and Development,2009,46(5):841-849.
[7]Zhuge Jian-wei,Wang Da-wei,Chen Yu,et al.A network anomaly detector based on the D-S evidence theory[J].Journal of Software,2006,17(3):463-471.
[8]Li Peng,Liu Si-feng.Interval-valued intuitionistic fuzzy numbers decision-making method based on grey incidence analysis and D-S theory of evidence[J].Acta Automatica Sinica,2011,37(8):993-998.
[9]Qiu Pei-yuan,Lu Feng,Zhang Heng-cai.Extracting traffic information from web texts with a D-S evidence theory based approach[C].Proceedings of the 21st International Conference on Geoinformatics,2013:1-5.
[10]Huang Shi-guo,Zhou Ming-quan,Geng Guo-hua.Angel invariant gabor-SVMfor insect recognition[J].Journal of Chinese Computer Systems,2010,31(1):143-146.
[11]Chen Kai-zhi,Le Cheng-pei,Zhong Shang-ping.Fusion method of distance metric learning and SVMfor image matching[J].Journal of Chinese Computer Systems,2015,36(6):1353-1357.
[12]Zhang Xiang,Xiao Xiao-ling,Xu Guang-you.Fuzzy support vector machine based on affinity among samples[J].Journal of Software,2006,17(5):951-958.
[13]Platt John C.Probabilistic outputs for support vector machines and comparisons to regularized likelihood methods[M].Alexander J.Smola.Advances in Large Margin Classifiers,Cambridge,MA:MIT Press,1999:1-11.
[14]Wang Guan,Xie Si-hong,Liu Bing,et al.Reviewgraph based online store reviewspammer[C].Proceedings of the 11th IEEE International Conference on Data Mining(ICDM),2011:1242-1247.
[15]Liang Dong-xu,Liu Xin-yue,Shen Hua.Detecting spam reviewers by combing reviewer feature and relationship[C].Proceedings of the International Conference on Informative&Cybernetics for Computational Social Systems(ICCSS),2014:102-107.
[16]Lu Yu-qing,Zhang Lei,Xiao Yu-dong,et al.Simultaneously detecting fake reviews and reviewspammers using factor graph model[C].Proceedings of the 5th ACMWeb Science Conference,2013:225-233.
[17]Rayana Shebuti,Akoglu Leman.Collective opinion spam detection bridging reviewnetworks and metadata[C].Proceedings of the21th ACMSIGKDD International Conference on Knowledge Discovery&Data Mining,Sydney,NSW,Australia,August 10-13,2015:985-994.
[18]Mukherjee Arjun,Kumar Abhinav,Liu Bing,et al.Spotting opinion spammers using behavioral footprints[C].Proceedings of the 19th ACMSIGKDD International Conference on Knowledge Discovery&Data Mining,2013:632-640.
[19]Xue Hao,Li Feng-jun,Seo Hyunjin,et al.Trust-Aware reviewspam detection[C].Proceedings of the 14th IEEE International Conference on Trust,Security and Privacy in Computing and Communications,2015,1:726-733.
[20]Ye Jun-ting,Kumar Santhosh,Akoglu Leman.Temporal opinion spam detection by multivariate indicative signals[C].Proceedings of the 10th International AAAI Conference on Web and Social Media,2016:743-746.
[21]Fei Ge-li,Mukherjee Arjun,Liu Bing,et al.Exploting burstiness in reviews for reviewspammer detection[C].Proceedings of the 7th International AAAI Conference on Weblogs and Social Media,2013:175-184.
[22]Xie Si-hong,Wang Guan,Lin Shu-yang,et al.Reviewspam detection via temporal pattern discovery[C].Proceedings of the 18th ACMSIGKDD International Conference on Knowledge Discovery&Data Mining,2012:823-831.
[23]Choo Euijin,Yu Ting,Chi Min.Detecting opinion spammer groups through community discovery and sentiment analysis[C].Proceedings of the IFIP Conference on Data&Applications Security&Privacy,2015,9149:170-187.
[24]Yang Min,Lu Zi-yu,Chen Xiao-jun,et al.Detecting reviewspammer groups[C].Proceedings of the 31st AAAI Conference on Artificial Intelligence,2017:5011-5012.
[25]Ye Jun-ting,Akoglu Leman.Discovering opinion spammer groups by network footprints[C].Proceedings of the Joint European Conference on Machine Learning&Knowledge Discovery in Databases,2015,9284:267-282.
[26]Salton Gerard.Experiments in automatic thesaurus construction for information retrieval[C].Proceedings of IFIP Congress,1971,71:115-123.
[27]Jacob Cohen.A coefficient of agreement for nominal scales[J].Educational&Psychological Measurement,1960,20(1):37-46.
[28]Kohavi Ron.A study of cross-validation and bootstrap for accuracy estimation and model selection[C].Proceedings of International Joint Conference on Artificial Intelligence(IJCAI),1995,14:1137-1143.
[29]Jarvelin Kalervo,Kekalainen Jaana.IR evaluation methods for retrieving highly relevant documents[C].Proceedings of the 29th international ACMSIGIR Conference on Research&Development in Information Retrieval,2000:41-48.
[1]彭岚,周启海,邱江涛.消费者在线评论有用性影响因素模型研究[J].计算机科学,2011,38(8):205-207.
[2]莫倩,杨珂.网络水军识别研究[J].软件学报,2014,25(7):1505-1526.
[4]段新生.证据理论与决策、人工智能[M].北京:中国人民大学出版社,1993.
[6]岳昆,刘惟一,王晓玲,等.一种基于不确定性因素叠加的W eb服务质量度量方法[J].计算机研究与发展,2009,46(5):841-849.
[7]诸葛建伟,王大为,陈昱,等.基于D-S证据理论的网络异常检测方法[J].软件学报,2006,17(3):463-471.
[8]李鹏,刘思峰.基于灰色关联分析和D-S证据理论的区间直觉模糊决策方法[J].自动化学报,2011,37(8):993-998.
[10]黄世国,周明全,耿国华.一种角度无关的Gabor-SVM昆虫识别[J].小型微型计算机系统,2010,31(1):143-146.
[11]陈开志,乐承沛,钟尚平.融合距离度量学习和SVM的图像匹配算法[J].小型微型计算机系统,2015,36(6):1353-1357.
[12]张翔,肖小玲,徐光祐.基于样本之间紧密度的模糊支持向量机方法[J].软件学报,2006,17(5):951-958.