简介:TheInternetnowisalarge-scaleplatformwithbigdata.Findingtruthfromahugedatasethasattractedextensiveattention,whichcanmaintainthequalityofdatacollectedbyusersandprovideuserswithaccurateandefficientdata.However,currenttruthfinderalgorithmsareunsatisfying,becauseoftheirlowaccuracyandcomplication.Thispaperproposesatruthfinderalgorithmbasedonentityattributes(TFAEA).Basedontheiterativecomputationofsourcereliabilityandfactaccuracy,TFAEAconsiderstheinteractivedegreeamongfactsandthedegreeofdependenceamongsources,tosimplifythetypicaltruthfinderalgorithms.Inordertoimprovetheaccuracyofthem,TFAEAcombinestheonewaytextsimilarityandthefactualconflicttocalculatethemutualsupportdegreeamongfacts.Furthermore,TFAEAutilizesthesymmetricsaturationofdatasourcestocalculatethedegreeofdependenceamongsources.TheexperimentalresultsshowthatTFAEAisnotonlymorestable,butalsomoreaccuratethanthetypicaltruthfinderalgorithms.