《基于IS+-树模型的频繁模式挖掘 Mining Frequent Patterns Based on IS+-Tree Model》.pdfVIP

11—1777／TP 计算机研究与发展 ISSN1000—1239／cN Joumalof Researchand Computer Development 基于IS t树模型的频繁模式挖掘 马海兵 张成洪 张 锦 胡运发 (复旦大学计算机与信息技术系 上海200433) (martin072l@163．com) PatternsBasedonIS1：TreeModel MiningFrequent Ma Jin，andHuYunfa Haibing，ZhangChenghong，Zhang n起d (．D日p口rf，咒已起￡0．厂CD7咒p“￡PrJ以如r优口fio靠丁kc^咒oz(喀y，f讹，zL肺it肥，百i砂，S^n以g矗ni200433) AbstractIS—treeisanovelmathematicalmodel hasbeen to presentedrecently，whichsuccessfuUyapplied full—textindexand intextdatabase．Inthis isextendedtodata andan storage paper，itsapplication mining is for basedonIS+_tree．The builds algorithmpresentedminingfrequentpatterns algorithmfrequent as does． ithasseveral oVertheFP—tree algorithmHowever， patternsdirectly，FP—growth advantages scansthetransactiondatabase is model．Firstly，it onlyonce．Secondly，theminingp南cessonly withonerOot canbe increments．The tree．Thirdly，IS+_treedynamicaUyupdatedby performancestudy showsthatthe is tooreven than isa all，IS+-tree algorithmefficiencyequal higherFP—growth．Above g∞d modeltoindextransaction itcan ontransactionswith database，andsupportquery highefficiency． words Key frequentpattern；Apriorialgorithm；FP—growthalgorithm；IS—tree；IS+_tree 摘要IS一树是一种新型的全文存储索引模型．提出一种基于扩展IS+_树模型的频繁模式挖掘算法．和 测试操作．然而它比FP一树模型具有更多的优点：只需扫描一遍事务库；挖掘任务只局部关联于一棵根 是，IS十_树模型同时是