欢迎来到天天文库
浏览记录
ID:34102426
大小:4.96 MB
页数:62页
时间:2019-03-03
《微博中影响力的分析》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、ABSTRACTWithtlledeVelopmentofW曲2.0technology,varioussocialmediahavespmnguplnrecentyears.AsapreVailingone,Microbloghasbeenanimponantwayinthedallymutualcommunicationsandinfonnationspreadamongpeople.DuetosomeunlquecharacteristicsMicroblogoWns,a铲owingn啪berofresearcheshaVepaidattentlonto1tand锄ongthese
2、researches,iIlfluenceisahottopic,includinginnuencemeaswementandinfluencemaximization.Eventhou曲someachievementshavebeenobtained,therearestillsomeproblems.Forex砌ple,amongalltheproposedalgo“山=tlstoapproximatetheinnuencemaXimization,somearetiIne.consumin2,somelnaccurateandsomeincludingtoomallyassumpt
3、ions.Regaurdingtheinnuencemeasure】ment,thereareindeedacertainnumberofnovelperspectivesdeservingexploration.Therefore,thispaperwillfocusontheinnuenceinMicroblogf.romboththetv旧topics.Accordingtothebidirectionali够ofuserinteractions,mispaper6rstestjmatestheInteractlVedegreebetweentwousersfromretweeti
4、ngstrength,commentingintensity,mentioningdens时andkeywordsimilar对thesefourfactofs.Then,consideringthatafollower’scontributionstotheinnuencesofhismerfolloweesvarvanddepend鲈eatlyontheinteractionsbetweenthem,anovelinnuencemeasurementalgoritllIllMBUserRankisputfonvard,whichisthebasisoftheinnuencemaxim
5、ization.Further,tobetterapproximatetheactivationprobabilityandreducethecomplexi劬amodi伍edalgorithmMBRankbasedonMBUserRankisgiven.WiththerallkingresuItsofMBRankonaUtheusersinthene帆ork,sometopra】:1l(edusersareselectedtoconstituteacandidateseedset.Asthisnewheuristicisintroduced。anovelinnuencemaximiza
6、tionalgorithmMB(heedyanditsimprovementMBCELFareproposed.Finally,acommonproblemindailylifewhetherthetop—krankedusersalwaysleadtotheinnuencemaximizationisdiscussedajldm11ingformecandidateseedsizeisillustrated。ThroughtheextensiVeexperimentsonTencentWeibodataset,thefollowjngconclusionsaremade.First,t
7、henewpr叩osedinnuencemeasurementalgoritl瑚MBUserRankisreaUyc01I】【petentingivingabetterrankingresultclosertorealiW.Second,combiningboththeadvarltagesofgreedy.basedandheuristic.basedalgorithns,MBCELFcanobtainagoodapproxima
此文档下载收益归作者所有