深度学习视角下基于多模态知识图谱的MOOC课程重构

Reconstruction of MOOC Courses based on Multimodal Knowledge Map from the Perspective of Deep Learning

文章以在线学习者的深度学习需求为出发点,首先提取来自不同平台的MOOC课程资源所蕴含的知识实体及其层级关系,构建以"知识实体—关系—知识实体"三元组为核心的多模态知识图谱。随后,文章通过多个相同内容的MOOC课程的资源融合,重构面向学习者深度学习需求的个性化MOOC课程。最后,文章依据在线学习者所设定的学习目标生成动态学习路径,为学习者提供优质课程资源推荐服务,解决MOOC课程的学习导航问题和优质课程资源的跨平台融合问题。实验表明,基于多模态知识图谱的MOOC课程重构能够降低MOOC课程的辍学率,提高在线学习者的学习成效和学习效率。

In consideration of the online learners’deep learning needs,this article firstly extracted the knowledge entities and their hierarchical relationships contained in MOOC course resources from different platforms,and then constructed a multimodal knowledge map with the“knowledge entity-relationship-knowledge entity”triplet as the core.Secondly,the personalized MOOC courses oriented to learners’deep learning needs were reconstructed through the resource integration of multiple MOOC courses with the same content.Finally,the dynamic learning path was generated according to the learning objectives set by online learners.It could recommend high-quality course resources to online learners,solve the problems of learning navigation of MOOC courses and cross-platform integration of high-quality course resources.The experimental results showed that the reconstruction of MOOC based on multimodal knowledge map could reduce the dropout ratio of MOOC courses,and improve the learning effectiveness and learning efficiency of online learners.