基于潛在語義索引的文本聚類技術(shù)研究.pdf

1、沈陽航空工業(yè)學(xué)院碩士學(xué)位論文基于潛在語義索引的文本聚類技術(shù)研究姓名：鄭偉申請(qǐng)學(xué)位級(jí)別：碩士專業(yè)：計(jì)算機(jī)應(yīng)用技術(shù)指導(dǎo)教師：蔡?hào)|風(fēng)；劉紀(jì)元20090307AbstractWiththedevelopmentofinformationtechnologyalargeamountofdocumentsourcesareneededtobeeffectivelyorganizedfortopicdiscoveryinformationretrie

2、val，etcTomeettheserequirements，thedocumentclusteringtechniqueemergesintime，whichisanimportantresearchtopicofnaturallanguageprocessingManyprogresseshavebeenmadeintheresearchofdocumentclusteringThenaturallanguagephenomenas

3、uchasagreatnumberofsynonymsandpolesemyexistindocumentclusteringLatentSemanticIndexing(LSI)isusedtodiscussandresolvethesephenomenainordertoimprovetheperformanceofdocumentclusteringinthisthesisSingularValueDecomposition(SV

4、D)technologyoftheLSItransformstheoriginaltermspacetothecorrespondingsmallerlatentsemanticspace，duringwhichthetermswithhighdocumentfrequencyintroducesomeunreasonabletermtransferrelationsthatinfluencethesimilaritybetweente

5、rmsandthesimilaritybetweendocumentsinthedocumentsetsThisthesisproposesafeatureoptimizetechnologyinlatentsemanticindexingbymakinguseofthetransferrelationoftermsinthedocumentsandbetweenthedocumentsindocumentsetsThismethodc

6、anchoosethetransferrelationsinlatentsemanticspace，andtheexperimentalresultsshowthatthismethodcanimprovetheperformanceofLSIeffectivelyIntheresearchofdocumentclusteringalgorithmclusteringalgorithmsbasedonpartitionaresensit

7、ivetotheinitialpointsandpronetobetrappedinlocaloptimizationThisthesisproposesamethodbasedonthecenterofsumfunctionoftheminimalsimilaritybyanalyzingthecharacterofinitialpointsinthethesisTheKdocumentsareselectedastheinitial

8、pointsofthedifferentcategoriesindocumentsetsandthesimilaritysumoftheseKdocumentsaretobethesmallestinthismethodSothismethodavoidssplittingthecategorywhichhasagreatofdocumentsintosmallcategoriesandtheinitialpointstobeborde