资源描述：
《Frequent subgraph discovery in dynamic networks_MLG10》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、FrequentSubgraphDiscoveryinDynamicNetworksBiancaWackersreutherPeterWackersreutherAnnahitaOswaldInstituteforInformaticsInstituteforInformaticsInstituteforInformaticsUniversityofMunichUniversityofMunichUniversityofMunichMunich,GermanyMunich,GermanyMunich,German
2、ywackersb@dbs.ifi.lmu.dewackersr@dbs.ifi.lmu.deoswald@dbs.ifi.lmu.deChristianBöhmKarstenM.BorgwardtInstituteforInformaticsMPIsforDevelopmentalUniversityofMunichBiologyandBiologicalMunich,GermanyCyberneticsboehm@dbs.ifi.lmu.deTübingen,Germanykarsten.borgwardt@tueb
3、ingen.mpg.deABSTRACTnenthastobetakenintoaccount,asinteractionsbetweenobjectshereusuallyoccurforacertainperiodoftimeonly.Inmanyapplicationdomains,graphsareutilizedtomodelTherefore,arealisticmodelhastoconsiderthatedgeswillentitiesandtheirrelationships,andgraphm
4、iningisimpor-beinsertedand/ordeletedovertime.Theresultingdatatanttodetectpatternswithintheserelationships.Whilethestructureiscalledadynamicgraph.majorityofrecentdataminingtechniquesdealwithstaticThesedynamicngraphsoccurinmanyreal-worldapplica-graphsthatdonotc
5、hangeovertime,recentyearshavewit-tions.InBiology,awide-spreadapproachistomodelinter-nessedtheadventofanincreasingnumberoftimeseriesactingproteinsasnetworks,whereeachvertexcorrespondsofgraphs.Inthispaper,wede neanovelframeworktotoaproteinandtwoverticesareconne
6、ctedbyanedgeifperformfrequentsubgraphdiscoveryindynamicnetworks.thecorrespondingproteinscanbind.Inaddition,furtherInparticular,weareconsideringdynamicgraphswithedgetechnologiesallowbiologiststomeasurethedistributionofinsertionsandedgedeletionsovertime.Existin
7、gsubgraphproteininteractionsatdi erenttimepoints.Hence,asso-miningalgorithmscanbeeasilyintegratedintoourframe-ciatingatimeseriesforeachproteinprovidesinterestingworktomakethemhandledynamicgraphs.Finally,aninsightsintothedynamicallychangingsystem.Insocialexten
8、siveexperimentalevaluationonalargereal-worldcasenetworkslikeFacebook,peoplecontacteachotheratspe-studycon rmsthepracticalfeasibilityofourapproach.ci ctimepointsandformvariouscomplexrelati
