欢迎来到天天文库
浏览记录
ID:36562350
大小:4.10 MB
页数:132页
时间:2019-05-12
《结构数据挖掘与处理的若干问题的研究》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、Y769599学校代码:10246学号:021021082後璺大学博士学位论文结构数据挖掘与处理的若干问题的研究院专姓系:信息科学与工程学院业:计算机软件与理论名:王晨指导教师:施伯乐教授完成日期:2005年4月20日复旦大学博士学位论文AbstractRecently,datamininganditsapplicationshavealreadycomeintomanydisciplinesandachievedplentifulfruitsindiversifiedfields,includingartificialintelligenceandmachin
2、elearning,database,patternrecognition,bioinformatics,neuralcomputing,andsoon./tnotonlyappealsscientistsbutalsocatchestheattentionfromgovernmentsandindustries.Thegovernments,industrialcommunities.andacademicfieldsaresokeenonmasteringdataminingtechniquesthattheyhaveinvestedalargedealof
3、moneyandenergyonthecorrespondingresearch.Therefore,theprogressofdata碍iningwillpromotethedevelopmentofscieneeandsociety.Withtheprogressofdataminingtechniques,moreandmorequestionshavebeenpresented.Thedemandofminingoncomplexdataisrisingnow.Expertshavepaidattentiontothesefieldsandtriedto
4、solvetheproblemsbyvirtueoftheexperienceofunstructureddatamininglikefrequentitemsetsmining.Inthispaper,IdotheresearchOilstructureddataminingandprocessing.Inthisdissertation,4problemsstandinginneedofsolutionsareinvestigated,whichincludesimprovingtheefficiencyofsemi—structureddatamining
5、,promotingthesealabilityofstructureddatamining,mininggraphdatawithconstraints,andindexinggraphdatabase.Themaincontributionsofthedissertationaresummarizedasfollows:Firstly,4algorithms,Chopper,XSpanner,ESMinerandISMiner,havebeenproposed.Thosealgorithmsminesfiequentinducedandembeddedsub
6、treesbyvirtueofmethodofpatterngrowflaandrightmostpathgrowthrespectively.ExperimentalresultsshowthatthealgorithmsperformbetterthanthosealgorithmspresentedagolikeTreeMinerandFREQT.Secondly,anovelgraphindexingstructureofADIisproposed.Itisembeddedintographminingalgorithmtoimprovethescala
7、bility.ExperimentalresultsshowthatADI—MineperformbetterthanotherslikegSpan,thebestgraphminingalgorithmbefore.Basedonit,IcontinuetopresenttheideasontransplantingtheADIindexingstructureintoothergraphminingalgorithmsforimprovingtheirefficiencyandscalability.j复旦火学博士学位论文Thirdly,thecommonc
8、onstraintsal
此文档下载收益归作者所有