社会网络的社团结构发现与动态特性分析-community structure discovery and dynamic characteristics analysis of social network.docx

AStudyofCommunityDetectionandDynamicPropertiesinSocialNetworksABSTRACTAllkindsofsocialnetworksplayamoreandmoreimportantroleinourdailylives.Byanalyzingusersbehaviorsinthesocialnetworks,wecangaintheinsightofthewayamessagepropagatinginthesocialnetworkandgettoknowhowtopromotenewideasandpreventspamsandrumorsfromepidemicspreading.Thestudyofthesocialnetworkcurrentlyistroubledbytheproblemthattheamountofthesocialnetworkdataistremendousandthemostvaluableinformationisconcealedinthewholedataset.Inthisworkwetacklethisproblemfromtwoaspects.Forthecommunitydetectionsection：wehaveproposedahierarchicaldiffusionmethodtodetectthecommunitystructurefromverylargesocialnetworks.Byusingthenetworkofcommunitiesinsteadofthenetworkofpeople,wecanreducethedimensionofthesocialnetworkgreatly.Ourcommunitydetectionalgorithmisbasedonthelocalstructure,soitsveryefficient.Testsonbothclassicalandsyntheticbenchmarksshowthatouralgorithmiscomparabletostate-of-the-artcommunitydetectionalgorithmsinbothcomputationalcomplexityandaccuracy.Forthedynamicpropertiessection：weinferusersinterestsfromtheirbehaviorsinsocialnetworks.Thenwepresentasupervisedlearningmethodforpersonalizedtweetsreorderingbasedonusersinterests.Twitterdisplaysthetweetsauserreceivedinareversedchronologicalorder,whichisnotalwaysthebestchoicebecausemanyinformativeorrelevanttweetsmightbefloodedordisplayedatthebottomduetosomenonsensebuzzes.Throughexploringarichsetofsocialandpersonalizedfeatures,wemodeltherelevanceoftweetsbyminimizingthepairwiselossofrelevantandnon-relevanttweets.Thetweetsarethenreorderedaccordingtothepredictedrelevancescoresfromthelearnedmodel.Experimentalresultswithrealtwitteruseractivitiesdemonstratedtheeffectivenessofourmethod.Thenewmethodachievedabove30%accuracygaincomparedwiththedefaultorderingintwitterbasedontime.Keywords:socialnetwork,diffusion,communitystructure,personalized,reorder目录社会网络的社团结构发现与动态特性研究I摘要IABSTRACTIII第一章绪论11.1社会网络研究的背景11.2社会网络研究的概述21.3社会网络的研究现状31.4章节安排3第二章社会网络的基本特性52.1社会网络的静态结构52.1.1“小世界”现象52.1.2幂律分布72.1.3连接强