基于图卷积神经网络的线上教育资源群组推荐方法  

Online Education Resource Group Recommendation Method Based on Graph Convolutional Neural Network

作　　者：周翰逊 张硕 张昉 孔壮壮 田思雨 王龙[1] 郭薇 毛艳娥 ZHOU Han-xun;ZHANG Shuo;ZHANG Fang;KONG Zhuang-zhuang;TIAN Si-yu;WANG Long;GUO Wei;MAO Yan-e(College of Information,Liaoning University,Shenyang 110036,China;Information Engineering Department,Liaoning Provincial College of Communications,Shenyang 110122,China;Information Engineering Department,Heze Vocational College,Heze 274002,China;Computer Academy,Shenyang Aerospace University,Shenyang 110136,China)

机构地区：[1]辽宁大学信息学院,辽宁沈阳110036 [2]辽宁省交通高等专科学校信息工程系,辽宁沈阳110122 [3]菏泽职业学院信息工程系,山东菏泽274002 [4]沈阳航空航天大学计算机学院,辽宁沈阳110136

出　　处：《辽宁大学学报（自然科学版）》2024年第2期169-178,共10页Journal of Liaoning University：Natural Sciences Edition

基　　金：辽宁省重点攻关项目(LZD20202);辽宁省教育厅项目(JYT19053);辽宁省自然科学基金项目(2020-MS-239);辽宁大学教学改革项目(2020YBXM127);沈阳航空航天大学研究生教育教学改革项目(YJSJG202211);沈阳航空航天大学校级教学改革项目(JG2023018);辽宁省普通高等教育本科教学改革研究项目(辽教办〔2021〕254号)。

摘　　要：线上教育资源推荐方法大部分是面向单个学习者的,然而不少学习活动多以多个学习者构成群组的形式进行.针对上述情况,首先,我们对群组成员间的社交网络进行建模,将学习者信任度、相似度和专业度以及衡量成员之间关系权重的个性度作为特征信息,然后通过图卷积神经网络学习聚合邻居信息得到个人偏好.其次,借助偏好融合策略得到群组的最终偏好.最后,按照项目评分,对线上学习群组进行教育资源推荐.实验结果表明,该方法有效提高了线上学习者群组的准确性和满意度.Most of the online educational resources recommendation methods are oriented to a single learner.However,many learning activities are carried out in the form of a group of multiple learners.In view of the above situation,firstly,we model the social network between group members,and use the learner′s trust,similarity and professionalism,as well as the personality degree that measures the weight of the relationship between members as the feature information,and then obtain personal preferences by learning to aggregate neighbor information through graph convolutional neural network.Secondly,the final preference of the group was obtained by using the preference fusion strategy.Finally,educational resources were recommended for online learning groups according to the project scores.Experimental results show that the proposed method effectively improves the accuracy and satisfaction of online learners.

关 键 词：群组推荐 成员关系 图卷积神经网络 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]