摘要
TheInternetnowisalarge-scaleplatformwithbigdata.Findingtruthfromahugedatasethasattractedextensiveattention,whichcanmaintainthequalityofdatacollectedbyusersandprovideuserswithaccurateandefficientdata.However,currenttruthfinderalgorithmsareunsatisfying,becauseoftheirlowaccuracyandcomplication.Thispaperproposesatruthfinderalgorithmbasedonentityattributes(TFAEA).Basedontheiterativecomputationofsourcereliabilityandfactaccuracy,TFAEAconsiderstheinteractivedegreeamongfactsandthedegreeofdependenceamongsources,tosimplifythetypicaltruthfinderalgorithms.Inordertoimprovetheaccuracyofthem,TFAEAcombinestheonewaytextsimilarityandthefactualconflicttocalculatethemutualsupportdegreeamongfacts.Furthermore,TFAEAutilizesthesymmetricsaturationofdatasourcestocalculatethedegreeofdependenceamongsources.TheexperimentalresultsshowthatTFAEAisnotonlymorestable,butalsomoreaccuratethanthetypicaltruthfinderalgorithms.
出版日期
2017年03月13日（中国期刊网平台首次上网日期，不代表论文的发表时间）
