多视角驱动的个性化课程推荐研究

打开文本图片集

中图分类号：TP319 文献标识码：A 文章编号：1006-8228（2025）11-07-06

Research on Multi-Perspective Driven Personalized Course Recommendations

DiaoLijuan，Wu Wei，Li Jizhao，JinXinying，Zhao Haibo，GengHao （North China Institute ofAerospace Engineering，LangFang，Hebei O65ooo，China）

Abstract：WithhewidespreadadoptionofMassveOpenOnineCourses（MOOC）platforms，learnersnowhaveaccesstoabundant educationalresources.Howeverwithinthisvastlandscapeofleaingmaterils，personalizedcourseselectionpresentscertain chalenges.TraditionalrecommendationmodelslacksuficientinterpretabilityForthisreason，thispaperproposesacourse recommendationmodel（CR-GCNMP）basedongraphconvolutionalnetworks（GCN）andmulti-viewmeta-paths，whichenhancesthe accuracyandrationalityofrecommendationsbyintegratingtheideaofpredictingpreferencesbasedonassociativeinteractionsin colaboratieodelsistaphonvolutioaletworkcusielypropagateslaeouseinteractioifoatiotoneateig orderemeddingsatptureollbatieifatioecodul-rspectivemeta-pathsetractedfrohulti-sioal atributecourseknowledgegraph.BidirectionalLSTMmodelsarethenemployedtoestablishseparatebidirectionalsemanticmodels forlearnerpreferencesandcoursesuitabilityFinallyanatentionmechanismdyamicallyaggregatesmeta-pathweightstogeerate interpretablerecommendationresults.ThispaperconductedexperimentsonthepublicdatasetXuetangX.Theexperimentalresults showthat CR-GCNMP outperforms the baseline models by 1.1% ， 7.2% ，and 8.1% in Precision@10，Recall@20，and NDCG@5， respectively，andefectivelyenhancesthelearningexperienceandlearningoutcomesoflearnersononlinelearningplatforms. Keywords：CollaborativeModel;CourseRecommendation;GraphNeural Network；Multi-Perspective Meta-Path

0引言

近年来，大规模开放在线课程（Massive OpenOnlineCourses，MOOC）已成为全球教育领域的重要资源，吸引了广大学习者的关注。（剩余7424字）