如何计算两个文档的相似度

LSI本质上识别了以文档为单位的second-orderco-ocurrence的单词并归入同一个子空间。因此：1）落在同一子空间的单词不一定是同义词，甚至不一定是在同情景下出现的单词，对于长篇文档尤其如是。2）LSI根本无法处理一词多义的单词（多义词），多义词会导致LSI效果变差。ApersistentmythinsearchmarketingcirclesisthatLSIgrantscontextuality;i.e.,termsoccurringinthesamecontext.Thisisnotalwaysthecase.ConsidertwodocumentsXandYandthreetermsA,BandCandwherein:AandBdonotco-occur.XmentionstermsAandCYmentionstermsBandC.:.A—C—BThecommondenominatorisC,sowedefinethisrelationasanin-transitco-occurrencesincebothAandBoccurwhileintransitwithC.Thisiscalledsecond-orderco-occurrenceandisaspecialcaseofhigh-orderco-occurrence.