文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > Efficient inverted lists and query algorithms for Structured Value Ranking in update-intens
侵权投诉

Efficient inverted lists and query algorithms for Structured Value Ranking in update-intens

We propose a new ranking paradigm for relational databases called Structured Value Ranking (SVR). SVR uses structured data values to score (rank) the results of keyword search queries over text columns. Our main contribution is a new family of inverted lis

Ef cientInvertedListsandQueryAlgorithmsforStructuredValueRanking

inUpdate–IntensiveRelationalDatabases

LinGuo

JayavelShanmugasundaram

CornellUniversityguolin,jai@cs.cornell.edu

Abstract

WeproposeanewrankingparadigmforrelationaldatabasescalledStructuredValueRanking(SVR).SVRusesstructureddatavaluestoscore(rank)theresultsofkey-wordsearchqueriesovertextcolumns.Ourmaincontri-butionisanewfamilyofinvertedlistindicesandassoci-atedqueryalgorithmsthatcansupportSVRef cientlyinupdate–intensivedatabases,wherethestructureddataval-ues(andhencethescoresofdocuments)changefrequently.Ourexperimentalresultsonrealandsyntheticdatasetsus-ingBerkeleyDBshowthatwecansupportSVRef cientlyinrelationaldatabases.

1Introduction

SQL/MM[23]isthestandardextensiontoSQLforpro-cessingtextdata.Currentrelationaldatabasesthatimple-mentSQL/MM[9,10]treatthetextmanagementcompo-nentasa“blackbox”andusetraditionalrankingtechniquessuchasTF–IDF[26].Suchrankingtechniques,however,wereoriginallydevelopedbytheIRcommunityforstand–alonedocumentcollectionsandoftendonotproducemean-ingfulresultsfortextstoredinrelationaldatabases.Asanillustration,considertherelationaldatabaseinFigure1,whichcontainsfragmentsofdatafromtheInternetArchive(doc.guandang.net)database,andtheSQL/MMquerybelow,whichreturnsthetop10movieswhosedescriptioncontainsthekeywords“goldengate”.UsingTF–IDF,thetwomoviesinthedatabasewillhaveaboutthesamescoreinthequeryresultbecausethekeywords“goldengate”occurthesamenumberoftimes(i.e.,once)inbothofthemandtheirtext eldlengthsareaboutthesame.Inotherwords,thetwomovieswillhavesimilarscoresdespitethefactthattheyhaveverydifferentpopularitiesbasedontheirreviewratingsandnumberofvisits/downloadsinthedatabase.SELECT*

FROM

Moviesm

ORDERBYscore(m.desc,"goldengate")FETCHTOP

10RESULTSONLY

KevinBeyerEugeneShekita

IBMAlmadenResearchCenterkbeyer,shekita@almaden.ibm.com

Inthispaper,weproposeStructuredValueRanking(SVR),whichusesthestructureddatavaluesrelatedtothetextdatainordertorankkeywordsearchqueryresults.Inourexample,usingSVR,wecanscoreamoviebasedonacombinationofits(a)averagereviewerratings,(b)num-berofvisitsbyusers,and(c)numberofdownloads.Thus,themovie“AmericanThrift”willbeclearlyrankedabove“AmateurFilm”sinceithasreceivedhigherratings,visitsanddownloads.OneofthecontributionsofthispaperisaSQL–basedframeworkforspecifyingandintegratingSVRinarelationaldatabase.Asystemadministratororanau-tomaticrankingtoolcanuseSQLqueriestospecifyhowtextcolumnsaretobescoredbasedonthevaluesofre-latedstructureddata(suchasaveragereviewerratings).OurframeworkalsoenablesrankingbasedonacombinationofSVRandTF–IDF.

ThemaincontributionofthispaperisanewfamilyofinvertedlistindicesandqueryalgorithmsthatcansupportSVRef cientlyinrelationaldatabases.Themainchal-lengeindevisingtheseindicesistoef cientlysupporttop–querieswhenthescoresofthetextcolumnschangefre-quentlyandpossiblydramatically.OurinitialmotivationforthisproblemaroseinthecontextoftheInternetArchivedatabase,wherethereviewerratings,numberofvisitsandnumberofdownloadswereupdatedoftenandsigni cantlychangedthescoresoftheassociatedtextcolumns(fre-quently,thesechangesaredueto“ ashcrowds”ontheInternet,whereanitemsuddenlygainspopularityduetosomeexternaleventsuchasanawardannouncement[2]).Fromtheuser’spointofview,theywouldstillliketoseethetoprankedresultsbasedonthelatestscoressothattheydonotmissimportantrecentinformation.Webelievethatthisisalsotrueforalargeclassofotherupdate–intensivedatabaseswhereSVRisapplicable,suchasstockdatabases(wherevolumeoftradecanbeusedtorankresults),on-lineauctionssuchase–bay(wheretimetocompletionandthecurrentbidcanbeusedtorankresults)andahostofotherweb–basedapplicationswherethepopularityofanitemsometimeschangesdramaticallyinaspanofafew

第1页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email