Research on Digital Forensics Based on Uyghur Web Text Classification

Abstract

Thispapermainlydiscussestheuseofmutualinformation(MI)andSupportVectorMachines(SVMs) forUyghurWebtextclassificationanddigitalforensicsprocessofwebtextcategorization:automatic classificationandidentification,conversionandpretreatmentofplaintextbasedonencodingfeatures ofvariousexistingUyghurWebdocumentsetc.,introducesthepre-paratoryworkforUyghurWeb textencoding.Focusingonthenon-Uyghurcharactersandstopwordsinthewebtextsfiltering, weputforwardaMulti-featureSpaceNormalizedMutualInformation(M-FNMI)algorithmand replaceMIbetweensinglefeatureandcategorywithmutualinformation(MI)betweeninputfeature combinationandcategorysoastoextractmoreaccuratefeaturewords;finally,weclassifyfeatures withsupportvectormachine(SVM)algorithm.Theexperimentalresultshowsthatthisschemehas ahighprecisionofclassificationandcanprovidecriterionfordigitalforensicswithspecificpurpose.