SPOC论坛中学习者兴趣主题建模及其与学习成效的关系研究

Research on Modeling of Learners’ Interest Topics and Its Relationship with Learning Outcomes in SPOC Forum

小规模限制性在线课程(Small Private Online Course,SPOC)论坛作为高校师生互动交流的重要场所,其产生的对话文本为教师了解学习者兴趣、实现智能化和个性化教学带来契机。研究提出了一种基于情感-主题时序建模的学习者兴趣主题检测与追踪方法,用于发现SPOC论坛中学习者兴趣主题与教学内容的相关性以及对学习成效的影响。研究结果表明:(1)与教学内容相关的兴趣主题对学习成效具有显著的正面效应,而娱乐灌水类兴趣主题对学习成效具有显著的负面效应;(2)在学期初期和末期,学习者兴趣主题与教学内容的语义相关度会显著上升,而中期则显著下降,即中期的"分心"现象最严重;(3)高成效组最先发起并参与了与教学内容相关的兴趣主题讨论,而低成效组的参与则较为滞后,且更倾向于表述与教学内容无关的内容。最后,通过对兴趣主题与学习成效的关系、兴趣主题对学习成效的影响以及不同成效群体在兴趣主题上的差异性与可预测性进行讨论,研究为互动教学中的精准学习干预以及智能导学方法的设计提供了相关建议。

The Small Private Online Course(SPOC) forum is an important space for interaction and communication between teachers and students in colleges and universities. The dialogue texts generated from the forum provide teachers with an insight into learners’ interests and bring opportunities for intelligent and personalized education. This study proposes a method of detecting and tracking the topics of learners’ interests based on temporal emotion-topic modeling, which is used to discover the correlation between the topics of learners’ interests and teaching content in SPOC forum and the impact on learners’ learning outcomes. The results indicate that, the topics of interest related to teaching content have a significant positive effect on learning outcomes, while the topics of interest related to entertainment have a significant negative effect on learning effects. At the beginning and the end of the semester, the semantic relevance between learners’ topics of interest and the teaching content will increase significantly, while at the mid-term stage, it decreases significantly. That is, the "distraction" is most serious. The high-efficiency group first initiates and participates in the discussion of the content-related topics of interest,while the low-efficiency group lags behind and tends to express unrelated topics. Finally, by discussing the relationship between the topics of interest and learning outcomes, the impact of topics of interest on learning outcomes as well as the differences and predictability of different groups in terms of topics of interest, this study provides some relevant suggestions for the precise learning interventions in interactive teaching and the design of intelligent tutoring methods.