面向智慧教育的自适应学习路径生成方法探究  

Research on Adaptive Learning Path Generation Method for Intelligent Education

作　　者：张晗[1,2] 赵润哲 芦培龙 张亚洲 姬莉霞[1,2] ZHANG Han;ZHAO Runzhe;LU Peilong;ZHANG Yazhou;JI Lixia(School of Cyber Science and Engineering,Zhengzhou University,Zhengzhou 450002,China;Zhengzhou Key Laboratory of Blockchain and Data Intelligence,Zhengzhou 450002,China)

机构地区：[1]郑州大学网络空间安全学院,河南郑州450002 [2]郑州市区块链与数据智能重点实验室,河南郑州450002

出　　处：《郑州大学学报（理学版）》2022年第6期59-65,共7页Journal of Zhengzhou University:Natural Science Edition

基　　金：河南省重大科技专项(201300210500);郑州市重大科技创新专项(2020CXZX0053)。

摘　　要：针对当前基于知识图谱的学习路径推荐算法仅使用单一的关系将学习对象连接起来,不能生成不同的学习路径来满足不同学习者的学习需求等问题,提出一种知识图谱与深度强化学习相结合的自适应学习路径生成方法。首先,构建出将细粒度知识元与学生认知能力相结合的知识图谱模型,并对知识单元之间的多种相互关系进行了预定义。然后,使用基于动态强化学习的路径生成方法,将DQN和DDQN通过不同权重整合入深度强化学习框架。最后,通过实验验证所提出的模型能够生成符合要求的个性化学习路径。Learning path generation algorithm based on knowledge graph attracted increasing attention of researchers.However,the current path recommendation algorithm based on knowledge graph only used a single relation to connect learning objects which could not generate different learning paths to meet the needs of different learners.To solve the problem,an adaptive learning path generation method combining knowledge graph and deep reinforcement learning was proposed.Firstly,a knowledge graph model combining fine-grained knowledge elements with students′cognitive level was proposed,and the relationships among knowledge elements were predefined.Then,a path generation method based on dynamic reinforcement learning was proposed.Deep Q network(DQN)and double deep Q network(DDQN)were integrated into the framework of deep reinforcement learning through different weights.Experimental re sults showed that the proposed model could generate personalized learning paths that meet the requirements.

关 键 词：自适应学习路径 知识图谱 智慧教育 深度强化学习 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]