基于改進(jìn)KNN的個(gè)性化郵件分類研究.pdf

1、重慶郵電大學(xué)碩士學(xué)位論文基于改進(jìn)KNN的個(gè)性化郵件分類研究姓名：鄧文韜申請(qǐng)學(xué)位級(jí)別：碩士專業(yè)：計(jì)算機(jī)軟件與理論指導(dǎo)教師：王國(guó)胤20110609AbstractWiththedevelopmentofInternet，E。mailhasspreadtoallpopulationsusingthenetworkasaconvenientandeconomicalwayofcommunication，butduetotheInterestso

2、fthedrivers，spareisalsoinparallelAvarietyofspampollutedthenetworkenvironment，andwastedlotsoftimeAnti—spareproblemhasbecomeimportantissuewhichalloftheworldconcernsBecausethesocialdivisionofworkandpersonalhobbiesisdifferen

3、t，thedemandofthepeopleforinformationgetmoreandmoreprofessional，andmoreandmorepersonalizedTheemergenceofpersonalizedservicetechnologysatisfiesuserpracticalneedsSothepersonalizedservicetechnologyisappliedtoEmailclassificat

4、ionisanewresearchdirectionInordertoachievepersonalizedservice，weneedtodesignanappropriateexpressiontocalculateuserinterestFinally，accordingtouserinterestofemail，wecanjudgewhetheranemailisspamornon—spamThejudgmentthatwhet

5、heranemailisspareornon。spammayvaryfrompersontopersonDifferentindividualscanhavetotallydifferentresponsestothesameemailFirstly，throughthereviewandembeddedstudyofthetheoryofpersonalizedserviceandEmailclassification，wehavep

6、resentedanewmethodofcomputinguserinterestbasedonusers’behaviorsRegardingsuchtraitsasirregularityofreceivingtheemailsandimbalancebetweenemailcategoryandthequantity，thethesisputsforwardapersonalizedimprovedKNNalgorithmoffi

7、lteringtheemailsImprovingthecountingmethodoninterestsandremovingtheinfluenceoftheusers’habitsontheestablishmentofinterestmodel，themethodperfectsKNNalgorithmandthusachievesbetterclassificationoftheemailsbasedontheresearch

8、astotheissuesofcategorydeviationExperimentsshowthatimprovedKNNfilteringmethodcouldhelptheusersclassifytheemailsaccordingtotheusers’interestsmuchmoreaccuratelyAtlast，basedontheimprovedKNNclassificationalgorithm，anEmailcla