高校SPOC讨论区的学习者话题演化分析研究

Research on the Analysis of Learners’Topic Evolution in SPOC Forums

SPOC讨论区中学习者讨论的话题具有时间转移、周期生存等特性,自动化地探索学习者话题演化分析已引起研究学者的广泛关注。以我国某高校云课堂平台中学习者发布的讨论帖为研究对象,首先基于文档时间信息统计学习者在同一学期不同时间单元内的发帖数量,其次在LDA话题模型的基础上,引入时间参数构建DTI-LDA(Deterministic Time Information-LDA)模型,最后分别采用先/后时间离散的分析方法,动态地追踪学习者关注话题的强度和语义内容的演化脉络。研究结果表明,学习者在讨论区中发帖的数量呈先缓慢上升后急剧下降趋势,并且学习者关注话题的强度及内容随时间呈现一定的演化规律。此外,可视化演化结果有助于教师整体把控学习者关注话题的发展轨迹,从而为实施有效的干预措施提供教学决策支持。

The topics discussed by learners in the SPOC forums have some characteristics,such as time transfer and periodic existence. Investigating the topic evolution of learners automatically has drawn an extensive attention from researchers. The post discussed from forums in cloud classroom platform of Central China Normal University,was used as the research object. Firstly,based on the time information of posts,this study counted the number of posts of the learners in different time units of the semester. Secondly,on the basis of the LDA,we incorporated an important time parameter to build the DTI-LDA model. Finally,the evolution of learner’s focused topic intensity and semantic content could be tracked dynamically by the first/last discrete time analysis method. The results indicated that the number of the learners’ posts in the forums slowed down first and then dropped sharply,and the intensity and content of learner’s focused topic showed a certain evolution rule. Besides,the visualization of the evolutionary results is helpful for teachers to monitor the development of learners’ concerned topics,thus providing instructional decision support for the implementation of precise interventions.