面向安全審計的約束序列挖掘研究.pdf

1、上海大學(xué)碩士學(xué)位論文面向安全審計的約束序列挖掘研究姓名：辛鴻亮申請學(xué)位級別：碩士專業(yè)：計算機應(yīng)用指導(dǎo)教師：歐陽為民20060101上洛太學(xué)碩士學(xué)位論文ABSTRACTDataflowofnetworkisinflatingsharplyandoperatingsystemisbecomingcomplicatedwhichcaHsethatthedataofsecurityauditalsoinereasesprogressivelyT

2、herearemuchimportantinformationshidingundertheastonishinglyincreasingdataWewanttotakefulladvantageofthedatabyanalyzingthemonahigherabstractlevelForanalyzingtheauditdatacomprehensivelyfastandaccuratelytheresearchonsecurit

3、yauditsystembasedondataminingbecomesarecenthotspotSequenceminingisanimportantproblemindataminingConsideringtheefficiencyandutilityofminin‰wewillusuallyintroducesomeconstraintsbasedonthefcaturesoftheconcreteareaHoweverthe

4、seconstraintsareverydispersiveinproposedmethodsandtherearetwoproblemsneedtesolve：firsttherearemanyconstraintsunincorporatedallthesame；sencondthereiSnosystemtofuseallkindsofconstraintsintooneminingprocessInthispaperweprop

5、oseanaudit一耐entedfastsequentialminingalgorithmSISC(SequenceminingwithStrictConstrants)basedonSPADE(SequentialPAtternDiscoveryusingEquivalenceclsses)wefusethetimegaDS，timewindowattributerelativeandattribute—schemaconstrai

6、ntsinonesequenceminingprocess，andinducethesequentialrulesfromconceptequivalenceclassesdirectlyExperimentsonareal—worldauditdatasetshowSISCoutperformsSPADEespeciallyremarkablywhentheaveragenumberofattributevaluesisfarlarg

7、erthanthenumberofattributesDataminin2istodiscovert11epotentialpatternsandrules，ifitCannotbeusedproperlyitwillendangertheprivacyandtheinformationSoprivacypreservingdataminingbecomesanewfoCUSintheresearchofdataminingInthis

8、PaDerweproposeanewtechniqueforprivacypreservingminingofsequentialpatternsItsstrategyiSasfollows：First，noisyeventsareaddedintoeachdatasequenceoftheoriginaldatabaseThen，analgorithmPP—Span(Privacy—PreservingSequentialpatter

9、nmining)isproposedtoreconstructthe矗equentsequencesfromthesenoise—addeddatasequencesExperimentalresultsindicatethatthismethodCanachievearatherhighlevelofaccuracyFurthermore，themethodCanutilizeanyexistingmethodofsequential

10、patternmining，SOitisofstrongadvantageinpracticeapplicationsSeqeunceminingwithconstraintsandprivacypreservingsequentialpatternminingarenotonlycompetentforthesecurityaudit，butalsothemedicalanalysis，scienceresearch，etcTheya