基于Flink的电子书推荐系统设计与实现

打开文本图片集

中图分类号：TP391.3 文献标志码：A 文章编号：2095-2945（2026）08-0145-05

Abstract：Withtherapiddevelopmentof theInteret，onlinee-bookresourceshaveexploded.Howtoeficientlyfindboks thatmeetone'sowninterestsinthevastoceanofE-bookshasbecomeanurgentproblemtobesolved.Personalizede-bk recommendationsystemprovidesanefectivewaytosolvethisproblembypredictinguserpreferencesandrecommending e-books thatmetuserinterests.However，thetraditionalcolaborativefilteringalgorithmhassomeproblemsinpracticalaplicationsuch aslowrecommendationeficiencyandinaccuraterecommendationresults，especiallyinthefaceof massivedataanddynamic changesinuserinterest.Inordertosolvetheseproblems，thispaperproposesanimprovedcolaborativeflteringalgorithm，and usesFlinkasaplatformtodevelopapersonalizede-bookrecommendationsystembasedonFlink.Themainresearchwork includes：thetimeattenuationfunctionisintroducedtogivediferentweightstousesbehavioratdiferenttimes，soasto capturethedynamicchangesofusers'interests，soastoimprovethequaltyofrecommendation;UsingFlink'spowerfulstreaming computingabilitywecanrealizethereal-timerecommendationofebooks，timelyrespondtousersreal-timeoperationbehavior andenhancetheuserexperience.Theexperimentalresultsshowthatthesystemcanefectivelyimprovetheaccuracyandrealtime of e-book recommendation，and better meet the growing personalized reading needs of users.

Keywords：recommendation system;collaborative filtering;time decay;Flink；e-book

互联网的快速发展让人们越来越依赖手机和电脑，使得数字化阅读得到快速发展，电子书作为闲暇时间的消遣读物，越来越受到广大人民的喜爱。（剩余5051字）