Abstract：Library book recommendation with CNN-FM deep learning approach PurposeTraditionallibrarybookrecommendationmethodsaremainlybasedonassociationrulesanduser profilesTheymayhelptolearnaboutstudentsinterestindifferenttypesofbooksegstudentsmajoringin scienceandengineeringtendtopaymoreattentiontocomputerbooksNeverthelessmostofthemstillneedto identifyusersinterestsaccuratelyTosolvetheproblemtheauthorsproposeanovelembedding-drivenmodel calledInFowhichreferstousersintrinsicinterestsandacademicpreferencestoprovidepersonalizedlibrary bookrecommendationsDesignmethodologyapproachTheauthorsanalyzethecharacteristicsandchallengesinreallibrary bookrecommendationsandthenproposeamethodconsideringfeatureinteractionsSpecificallytheauthors leveragetheattentionunittoextractstudentspreferencesfordifferentcategoriesofbooksfromtheir borrowinghistoryafterwhichwefeedtheunitintotheFactorizationMachinewithothercontext-aware featurestolearnstudentshybridinterestsTheauthorsemployaconvolutionneuralnetworktoextracthighordercorrelationsamongfeaturemapswhichareobtainedbytheouterproductbetweenfeatureembeddingsFindingsTheauthorsevaluatethemodelbyconductingexperimentsonareal-worlddatasetinone universityTheresultsshowthatthemodeloutperformsotherstate-of-the-artmethodsintermsoftwometrics calledRecallandNDCGResearchlimitationsimplicationsItrequiresaspecificdatasizetopreventoverfittingduringmodel trainingandtheproposedmethodmayfacetheuseritemcold-startchallengePracticalimplicationsTheembedding-drivenbookrecommendationmodelcouldbeappliedinreal librariestoprovidevaluablerecommendationsbasedonreaderspreferencesOriginalityvalueTheproposedmethodisapracticalembedding-drivenmodelthataccuratelycaptures diverseuserpreferencesLibrary AutomationThrough the method of automatic learning and recommendationbook recommendations are more accurate and personalizedConsulting ServiceProvides valuable book recommendations based on readerspreferencesRetrieval and IndexingThrough studentsbook history and other characteristicsprecise books are recommended Pattern recognitionidentify their interests by identifying and learning studentsborrowing historyMachine LearningWhen mentioning Factorization Machine and feature interactionthe use of machine learning is displayedDeep LearningThe article refers to the high-level correlation between the use of convolutional neural networks to extract featuresData MiningLearn and extract their interests through studentsbook history and other features