Filtering Spam in Social Tagging System with Dynamic Behavior Analysis.pdf

《Filtering Spam in Social Tagging System with Dynamic Behavior Analysis.pdf》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、2009AdvancesinSocialNetworkAnalysisandMiningFilteringSpaminSocialTaggingSystemwithDynamicBehaviorAnalysisBoLiu,EnnanZhai,HuipingSun,YeluChen,ZhongChenSchoolofSoftwareandMicroelectronics,PekingUniversity,Beijing,ChinaEmail:{lbo,enzhai,ylchen}@pku.edu.cn,{sunhp,chen}@ss.pku.edu.cnAbstract—Spam

2、insocialtaggingsystemsintroducedbythetagsmoreaccurately.Toelaborateourapproach,wefirstsomemaliciousparticipantshasbecomeaseriousproblemformodelbothofactionsandbehaviorsofusersandpresentitsglobalpopularizing.Somestudieswhichcanbededucedthedefinitionsofdifferentusersandbehaviors.Then,thetostatic

3、userdataanalysishavebeenpresentedtocombatprobabilitywhetherthetagisspamisassociatedwiththetagspam,buteithertheydonotgiveanexactevaluationorthealgorithms’performancesarenotgoodenough.Inprobabilityvalueoftheoccurrenceoftheabnormalbehavior.thispaper,weproposedanovelmethodbasedonanalysisBasedont

4、hebehaviormodel,analgorithmthatanalyzesofdynamicuserbehaviordataforthenotionthatusers’thebehaviorprocessinwhichusersavestheresourceandbehaviorsinsocialtaggingsystemcanreflectthequalityoftagschangesthetagisgiven.Experimentalresultsshowthatmoreaccurately.Throughmodelingthedifferentcategoriesofo

5、urapproachcanachievebetterperformancethanbothparticipants’behaviors,weextracttag-associatedactionswhichcanbeusedtoestimatewhethertagisspam,andthenpresentthepopularsortingusedinDel.icio.usandthecoincidenceouralgorithmthatcanfilterthetagspamintheresultsofalgorithmproposedin[4].socialsearch.Thee

6、xperimentresultsshowthatourmethodThecontributionsofthispaperinclude:indeedoutperformstheexistingmethodsbasedonstaticdata1)Wedefendthetagspamthroughminingtheuserandeffectivelydefendsagainstthetagspaminvariousspamattacks.behaviordataasaricherdatasource;2)User’sbehaviorhasbeenmodeledtoassociate

7、theKeywords-socialtaggingsystem;tagspam;abnormalbehav-probabilitywhetheratagisthespamtagwiththeior;probability;posting;probabilityvalueoftheoccurrenceoftheabnormalbehavior;I.INTRODUCTION3)Ananti-spamalgorithmbasedonbehaviordatahasTherapidgrowthinth