集值信息系統(tǒng)的知識發(fā)現(xiàn)與屬性約簡研究.pdf

1、西南交通大學(xué)博士學(xué)位論文集值信息系統(tǒng)的知識發(fā)現(xiàn)與屬性約簡研究姓名：陳子春申請學(xué)位級別：博士專業(yè)：交通信息工程及控制指導(dǎo)教師：秦克云201103第1I頁西南交通大學(xué)博士研究生學(xué)位論文AbstractRoughsettheoryisamathematicaltoolindataanalysisandCanbeeffectivelyusedtohandleincomplete，impreciseanduncertaininformationT

2、hetheoryhasbeensuccessfullyappliedtovariousfieldssuchasdatamining，machineleamingandpatternreeognitionOneimportantapplicationofroughsetistheknowledgediscoveryandknowledgereductionininformationsystemSetvaluedinformationsys

3、temisageneralizedmodelofsinglevaluedinformationsystem；itcallbeusedtocharacterizeuncertaininformationandinsufficientinformationWithroughsettheoryasthetool，andtherelatedoutcomesofresearchesasthebasis，thecurrentdissertation

4、systematicallystudiesknowledgeacquisitionandattributesreductionforsetvaluedinformationandsetvalueddecisiontableTheprimarycontributionscanbesummarizedasfollows：1Theknowledgeacquisitionandattributesreductionofset—valuedinf

5、ormationsystembasedonthevariableprecisiontolerancerelationarestudiedByintroducingavariableprecisiontolerancerelationtoset—valuedinformationsystem，theattributereductionunderacertainsimilarlevelareproposed；theinfluenceofsi

6、milarlevelontheattributereductionisanalyzed；theextractionofgeneralizeddecisionrulesandoptimalgeneralizeddecisionrulesfromasetvalueddecisiontablearediscussedFurthermore，inordertosimplifytheknowledgerepresentation，thedisse

7、rtationalsostudiestherelativereductionandtheassignmentreductionofsetvalueddecisiontablebasedonthevariableprecisiontolerancerelation，andgivessomeequivalentconditionsandtheimportantpropertiesofthesereductions2Basedonthemax

8、imumvariableprecisiontoleranceclasses，thedissertationstudiestheknowledgeacquisitionandattributesreductionofset—valuedinformationsystemWiththemaximumvariableprecisiontoleranceclassesasbasicsets，theattributereductionandrel

9、ativereductionarediscussed，theinherentrelationshipbetweentheattributereductionbasedonthevariableprecisiontolerancerelationaregivenFurthermore，thedissertationdiscussesthegeneralizeddecisionrulesandoptimalgeneralizeddecisi

10、onrulesoftheset—valueddecisiontablebasedonthemaximumvariableprecisiontoleranceclasses3TheknowledgeacquisitionandattributesreductionofsetvaluedinformationsystembasedonneighborhoodrelationsarestudiedThroughtheintroductiono

11、fneighborhoodrelations，theneighborhoodreductionofsetvaluedinformationsystemisproposed，thedefinitionandextractionmethodofneighborhooddecisionrulesinsetvalueddecisiontablearegivenInordertoderiveoptimalneighborhooddecisionr