基于lucene的全文检索系统设计与实现

《基于lucene的全文检索系统设计与实现》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、华中科技大学硕士学位论文AbstractCloudAVintelligentterminalisaintegrationproductoftelephonenetwork,TVnetworkandInternet,whichcombinespersonalmediacontentandtheInternet.ItmakesallthemediacontentcanbewatchedthroughtheHDpanelTV.TherearevastamountsofaudioandvideoresourcesinCloudA

2、VIntelligentTerminal.Usersneedefficientsearchingtoolstofindinformationtheyfocusonquicklyandprecisely.Eventhoughwehavematurefull-textsearchengineslikeBaidu,theintelligentterminalasaseparatesystemcan’tusethemdirectly.So,weneedtodesignanefficientfull-textretrievalsy

3、stem.Luceneisamatureandefficienttoolintheopensourceframeworksoffull-textretrievalsystem.Itisatoolkitoffull-textsearchenginedevelopedbyJava.WecanprocesssecondarydevelopmentwiththehelpoftheinterfacesprovidedbyLuceneinordertocompletefull-textretrievalsystemsofvariou

4、sspecificpurposes.Inthissystem,theobjectoftextanalysisismainlyfocusonChinese,sotheChinesewordsegmentationtechnologybecomesakeypoint.AlthoughLucenesupportsChinesewordsegmentation,themethodistoosimpleandmechanical.AndLucenehasthedefaultsortingalgorithm,buttheorderi

5、salwaysnotagreewiththefacts.So,usingLucenedirectlycannotmeettheactualneeds,wecantakeadvantageofLuceneonlyafterextendingandimprovingit.Wewilldesignandimplementafull-textretrievalsystembasedonLucenetosearchtheaudioandvideodataintheintelligentterminal.WecomeupwithGM

6、Malgorithm,GMMlooksforthelongestwordintheglobalrangeusingtheprincipleoftheglobalmaximummatchingforthepurposeofimprovingwordsegmentationaccuracy.Meanwhile,thesortingalgorithmofthereturnresultshasbeenimprovedthroughestablishinganewformulawhichconsideringvariousfact

7、orsliketimesandpositionstheII华中科技大学硕士学位论文keywordappears.TheresultsshowthattheimprovedGMMalgorithmhasgottenbetterwordsegmentationeffect,thesortingofthereturnsismoreinlinewithuser’srequirements.Atthesametime,therecallratioandtheprecisionratioofthefull-textretrieval

8、systemmaintainatahighlevel.Andthetimeofwaitingforqueryresultsiswithinacceptablelimits.Alltheaboveprovesthatthesystemmeetstheactualdemand.KeyWords：Full-textRetr