文献推荐系统：提高信息检索效率之途

？traditionalinformationretrieval(ir)systemshavelimitationsinimprovingsearchperformanceintoday’sinformationenvironment.thehighrecallandpoorprecisionoftraditionalirsystemsareonlyasgoodaswiththeaccuracyofsearchquery,whichis,however,usuallydifficultfortheusertoconstruct.itisalsotime-consumingfortheusertoevaluateeachsearchresult.therecommendationtechniqueshavingbeendevelopedsincetheearly1990shelpsolvetheproblemsthattraditionalirsystemshave.thispaperexplainsthebasicprocessandmajorelementsofdocumentrecommendersystems,especiallythetworecommendationtechniquesofcontent-basedfilteringandcollaborativefiltering.alsodiscussedaretheevaluationissueandtheproblemsthatcurrentdocumentrecommendersystemsarefacing,whichneedtobetakenintoaccountinfuturesystemdesigns.traditionalinformationretrieval(ir)systemshavelimitationsinimprovingsearchperformanceintoday’sinformationenvironment.thehighrecallandpoorprecisionoftraditionalirsystemsareonlyasgoodaswiththeaccuracyofsearchquery,whichis,however,usuallydifficultfortheusertoconstruct.itisalsotime-consumingfortheusertoevaluateeachsearchresult.therecommendationtechniqueshavingbeendevelopedsincetheearly1990shelpsolvetheproblemsthattraditionalirsystemshave.thispaperexplainsthebasicprocessandmajorelementsofdocumentrecommendersystems,especiallythetworecommendationtechniquesofcontent-basedfilteringandcollaborativefiltering.alsodiscussedaretheevaluationissueandtheproblemsthatcurrentdocumentrecommendersystemsarefacing,whichneedtobetakenintoaccountinfuturesystemdesigns.