资源描述：
《inside pagerank.pdf》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、InsidePageRankMONICABIANCHINI,MARCOGORI,andFRANCOSCARSELLIUniversityofSienaAlthoughtheinterestofaWebpageisstrictlyrelatedtoitscontentandtothesubjectivereaders’culturalbackground,ameasureofthepageauthoritycanbeprovidedthatonlydependsonthetopologicalstructureoftheWe
2、b.PageRankisanoticeablewaytoattachascoretoWebpagesonthebasisoftheWebconnectivity.Inthisarticle,welookinsidePageRanktodiscloseitsfundamentalpropertiesconcerningstability,complexityofcomputationalscheme,andcriticalroleofparametersinvolvedinthecomputation.Moreover,we
3、introduceacircuitanalysisthatallowsustounderstandthedistributionofthepagescore,thewaydifferentWebcommunitiesinteracteachother,theroleofdanglingpages(pageswithnooutlinks),andthesecretsforpromotionofWebpages.CategoriesandSubjectDescriptors:F.2.2[AnalysisofAlgorithms
4、andProblemComplexity]:NonnumericalAlgorithmsandProblems—Computationsondiscretestructures;H.3.3[Informa-tionStorageandRetrieval]:InformationSearchandRetrieval—Informationfiltering;H.5.4[InformationInterfacesandPresentation]:Hypertext/Hypermedia—NavigationGeneralTerm
5、s:AlgorithmsAdditionalKeyWordsandPhrases:Informationretrieval,Markovchains,PageRank,searchen-gines,searchingtheWeb,Webpagescoring1.INTRODUCTIONMostcommonlyusedscoringalgorithmsontheWeb,directlyderivedfromInfor-mationRetrieval(IR),employsimilaritymeasures,basedonafl
6、at,vector-spacemodelofeachpage.Unfortunately,thosemethodshavesomelimitationsanddrawbackswhenappliedtosearchingtheWeb,sincetheyonlytakeintoaccountthepagecontentandneglectthegraphicalstructureoftheWeb.Moreover,theseapproachesarepronetobecheated,sothatpagescanbehighl
7、yrankediftheycontainirrelevantbutpopularwords,appropriatelylocatedinthepage(e.g.,inthetitle).ThisphenomenonisusuallyreferredtoassearchenginepersuasionorWebspamming[Marchiori1997;Pringleetal.1998].ThisresearchwaspartiallysupportedbytheItalianMinistryofEducationunde
8、rgrantCOFIN2000.Authors’address:DipartimentodiIngegneriadell’Informazione,UniversityofSiena,ViaRoma,56–53100,Siena,Italy;email:{monica,marco,franco}@ing
