资源描述：
《Summarization and Matching of Density-Based Clusters in Streaming Environments 》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。
1、SummarizationandMatchingofDensity•BasedClustersin∗StreamingEnvironments†DiYangElkeA.RundensteinerMatthewO.WardOracleCorporationWorcesterPolytechnicInstituteWorcesterPolytechnicInstitute1OracleDrive100InstituteRoad100InstituteRoadNashua,NH,USAWorcester,MA,USAWorcester,MA,USAdi.yang@or
2、acle.comrundenst@cs.wpi.edumatt@cs.wpi.eduABSTRACTpatternminingsystemdoesnotonlyneedtobeequippedwithhighlyefficientpatternextractionalgorithms,butmoreDensity-basedclusterminingisknowntoserveabroadrangeimportantly,itmustalsoprovideeffectivepatternanalysisofapplicationsrangingfromstocktra
3、deanalysistomovingsupport,asmotivatedbelow:objectmonitoring.Althoughmethodsforefficientextrac-1)Patternfeatureabstraction.Thekeyfeaturesoftionofdensity-basedclustershavebeenstudiedinthelit-detectedpatternsmaybecomplexandthusmaynotbeerature,theproblemofsummarizingandmatchingofsucheasily
4、comprehensibleforhumananalystswithoutanalyticalclusterswitharbitraryshapesandcomplexclusterstruc-assistance.Forexample,inreal-timetrafficmonitoring,aturesremainsunsolved.Therefore,thegoalofourworkisclusterrepresentingacongestionareainthetrafficofBeijingtoextendthestate-of-artofdensity-ba
5、sedclustermininginmaybecomposedof10Korevenmorevehiclesandmaystreamsfromclusterextractiononlytonowalsosupport2spreadtoover10km.Bysimplylookingattheinformationanalysisandmanagementoftheextractedclusters.Ouraboutindividualclustermembers(vehicles),suchastheirworksolvesthreemajortechnical
6、challenges.First,wepro-positionsandmovingspeed,ananalystmaynotbeabletoposeanovelmulti-resolutionclustersummarizationmethod,identifythekeyfeaturesofthisclusterinrealtime,suchascalledSkeletalGridSummarization(SGS),whichcaptureswhereisthekeybottleneckcausingthecongestion.thekeyfeatureso
7、fdensity-basedclusters,coveringboth2)Patterncompression.Somepatternsneedtobetheirexternalshapeandinternalclusterstructures.Second,keptforlong-termanalysis,yetkeepingthefullrepresen-inordertosummarizetheextractedclustersinreal-time,wetationofthecomplexpatternstendstobeimpracticalinpre
8、sentanintegr
