1、Y930401學(xué)號(hào)：—03610—0422索引號(hào)；——藹弘≯塋擊哮題目：壘些昱控煎堡釜盛生鮑墨董隧亟豎墼墼壺法塹窒作者：鲞塑指導(dǎo)教師：堂撾壘專業(yè)技術(shù)職務(wù)：整蕉學(xué)科(專業(yè))；籃窒室墼劍造墨焦答辯日期學(xué)位授予日期：——二零零六年四月西安陌北下業(yè)人學(xué)壩I：學(xué)位論文AbslraclAbstractSemanticbasedenterpriseheterogeneousdatabaseintegrationneedstojudgewhichatt

2、ributessharethecorrespondingsemanticsHavingcomparedtheapproachesinandoutofourcountry’weproposeamultistrategyapproachforheterogeneousdatasemanticmatchingThisapproachcombinesseveralstrategiestoinsuretheeffectivenessofanaly

3、sisbyimportingtheholisticattributeinformationTheresearchesthatwefoCUSareasfollows：1)Multistrategysemanticmatchingapproachforheterogeneousdata：therealsemanticofattributeshouldbeexpressedbyholisticattributeinformationBycom

4、biningseveralstrategiestoanalyzetheholisticattributeinformation，Wig：caninsuretheeffectivenessofanalysis21StrategyA—NamematcherbasedonWordNet：Byimportingsemanticdictionary：WordNet，weproposeanapproachforcomputingsemanticsi

5、milaritybasedonWordNet3)Specificdictionaryexpression：InordertomakeuptheflawofWordNet，webuildaspecificdictionaryteinforceeffectivenessofstrategyA4)StrategyBSemanticclassifierbasedonSOMneuralnetwork：Bytranslatingschemainfo

6、rmationanddatavalueofattributesintoinputvectorstoimplementtheclassificationofattributesWeproposeamethodtoimproveinitializationofwei曲tbycomputingthemetadata’Ssensitivity5)Mergingmethodformultistrategysemanticmatchingresul

7、ts：ByusingstrategyAandB，wegettwosemanticmatchingsetsWeproposeabi—directionalrevisionmethodformergingmulti—strategyresultsandgetgoodjugement6)ParsetechnologyforXMLdocument：WeneedtousetheattributeinformationstoredintheXMLd