简介:Weappliedthedecisiontreealgorithmtolearnassociationrulesbetweenwebpage’scategory(pornographicornormal)andthecriticalfeatures.Basedontheserules,weproposedanefficientmethodoffilteringpornographicwebpageswiththefollowingmajoradvantages:1)aweightedwindow-basedtechniquewasproposedtoestimatefortheconditionofconceptdriftforthekeywordsfoundrecentlyinpornographicwebpages;2)checkingonlycontextsofwebpageswithoutscanningpictures;3)anincrementallearningmechanismwasdesignedtoincrementallyupdatethepornographickeyworddatabase.
简介:Givenanon-equidistantsequenceoranequidistantserieswithoneormoreoutliers,agreyinterpolationapproachconsideringthetimelagsisestablishedforproducingthemissingdataorcorrectingtheabnormalvalues.Toaccomplishthis,anewgreyincidencemodel,calledthegreydynamicincidencemodelGDIM(t),isconstructedfordeterminingwhetherthefactorsareeffectivetotheknownfactorandwhatthetimelagisbetweenausefulfactorandthespecifiedsequence.BasedontheresultsoftheGDIM(t)model,twoprogrammingproblemsaredesignedtoobtaintheupperandlowerboundsoftheunknownorabnormalvalueswhichareregardedasgreynumbers.Thesolutionsbasedontheparticleswarmoptimization(PSO)forthenonlinearprogrammingproblemsaregiven.Toexplainhowitcanbeusedinpractice,thisnewgreyinterpolationapproachisappliedtocorrectanabnormalvalueinthesequenceofanagricultureenvironmentproblem.