欢迎来到天天文库
浏览记录
ID:36354045
大小:3.38 MB
页数:107页
时间:2019-05-10
《基于有向项集图的关联规则挖掘算法研究与应用》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、天津大学博士学位论文基于有向项集图的关联规则挖掘算法研究与应用姓名:温磊申请学位级别:博士专业:管理科学与工程指导教师:李敏强20031201AbstractDalaminingwhichisalsoreferredasknowledgediscoveryindatabases,meansaprocessoffindingnontrivial,extractionofimplicit,perviousunknownandpotentialusefulinformationfromdataindatabase.Associationrulesmi
2、ningasailimportantfieldofdataminingdiscoverinterestingrelationshipsamongattributesinthosedata.Byreadingtheliteraturedomesticandabroad,Weresearchsomeproblemofassociationrulesminingalgorithms,themaincontextsandinnovationsareshowedasfoilow:1.Wediscusstherelationshipbetweenlatti
3、cetheory,formalconceptanalysisandassociationrulesminingandintroducaseriesofdefinitionandpropertyofassociationrulesmining.2,AnewfrequentitemselminingalgorithmsbasedDDDirectedJlemsetGraph(DISG)isintroduced.BystoringinformationoffrequentitemsetinDISG.Theproblemofdiscoveringthef
4、requentitemsetfromdatabaseistransformedintothesearchproblemofDISG.3.AnewmaximalfrequentitemsetminingalgorithmsbasedonDISGisintroducedtodiscoverthelongfrequentpattern.Byusingdepth-firststrategy,thealgorithmsprunethesearchingspacebycomputingthefrequentextensionsetofitemsetandd
5、iscoverallthemaximaJfrequentitemselefficiently.4.AnewalgorithmsofminingfrequentcloseditemselbasedonDISGisinlxoduced.Byusingdepth-firststrategythealgorithmsprunethesearchingspacebyjudgingthepropertyof台equeⅡtctosedseedsetanddiscoverallthefrequerttctoseitemselefficiently.5.Them
6、iningalgorithmsofincrementalupdatefrequentitemset,incrementalupdatemaximalfrequentitemsetandincrementalupdatefrequentcloseditemsetaredesignedbasedonDISG,.Thesealgorithmsc肋efficientlyutilizetheresu]lminedanddiscovertheupdatedfrequentitemsetefficiently.Thealgorithmsproposedint
7、hispaperistestedbyusingthelargescaledenseda'.asetwhicha11showgoodperformances.Wemakeanappticat{oneX嘲,';IAep.t.撕血thedatasetofpowerstationandachievesomevaluableinformation.Keyword:DataMioing,Directed[temsetGraph,AssociationRule,FrequentRemset,MaximalFrequentltemset.FrequentClo
8、sed]temset,IncrementalUpdateMining独创性声明本人声明所呈交的学位论文是本人在导师指导下进行的研究工作和取得的研究成果
此文档下载收益归作者所有