资源描述：
《a density based algorithm for discovering clusters in large spatial databsdes with noise 》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、PublishedinProceedingsof2ndInternationalConferenceonKnowledgeDiscoveryandDataMining(KDD-96)ADensity-BasedAlgorithmforDiscoveringClustersinLargeSpatialDatabaseswithNoiseMartinEster,Hans-PeterKriegel,JörgSander,XiaoweiXuInstituteforComputerScience,Univer
2、sityofMunichOettingenstr.67,D-80538München,Germany{ester
3、kriegel
4、sander
5、xwxu}@informatik.uni-muenchen.deAbstractareoftennotknowninadvancewhendealingwithlargeClusteringalgorithmsareattractiveforthetaskofclassiden-databases.tificationinspatialdatabases.Ho
6、wever,theapplicationto(2)Discoveryofclusterswitharbitraryshape,becausethelargespatialdatabasesrisesthefollowingrequirementsforshapeofclustersinspatialdatabasesmaybespherical,clusteringalgorithms:minimalrequirementsofdomaindrawn-out,linear,elongatedetc.
7、knowledgetodeterminetheinputparameters,discoveryof(3)Goodefficiencyonlargedatabases,i.e.ondatabasesofclusterswitharbitraryshapeandgoodefficiencyonlargeda-significantlymorethanjustafewthousandobjects.tabases.Thewell-knownclusteringalgorithmsoffernosolu-tio
8、ntothecombinationoftheserequirements.Inthispaper,Thewell-knownclusteringalgorithmsoffernosolutiontowepresentthenewclusteringalgorithmDBSCANrelyingonthecombinationoftheserequirements.Inthispaper,weadensity-basednotionofclusterswhichisdesignedtodis-prese
9、ntthenewclusteringalgorithmDBSCAN.Itrequirescoverclustersofarbitraryshape.DBSCANrequiresonlyoneonlyoneinputparameterandsupportstheuserindetermin-inputparameterandsupportstheuserindetermininganap-inganappropriatevalueforit.Itdiscoversclustersofarbi-prop
10、riatevalueforit.Weperformedanexperimentalevalua-traryshape.Finally,DBSCANisefficientevenforlargespa-tionoftheeffectivenessandefficiencyofDBSCANusingtialdatabases.Therestofthepaperisorganizedasfollows.syntheticdataandrealdataoftheSEQUOIA2000bench-Wediscus
11、sclusteringalgorithmsinsection2evaluatingmark.Theresultsofourexperimentsdemonstratethat(1)DBSCANissignificantlymoreeffectiveindiscoveringclus-themaccordingtotheaboverequirements.Insection3,wetersofarbitraryshapethanthewell-knownalgorithm
