现代远距离教育

同步在线讨论是在线协作学习中一种用于促进学习者知识建构的重要手段。评价同步在线讨论质量有助于学习者和教学者及时发现学习问题，并进行干预和调节。然而，当前质量评价研究多在异步在线讨论情景下开展，且在评价维度、评价粒度与评价方法上有待向多元化视角拓展。因此，首先面向同步在线讨论情景提出了包含认知、行为、情感及社交四个评价维度和个体、小组与班级三个评价粒度的质量评价框架，并给出了基于在线讨论文本的评价指标自动化计算与融合方法。为进一步发挥评价的促学作用，随后在评价框架的基础上构建了多维多粒度分析模型，阐述了评价工作的主要流程，强调通过使用不同的学习分析方法来分析个体、小组、班级在四个评价维度上的特点与差异，实现对讨论质量的深度评价。最后，将其应用在某师范高校三个班级的同步在线讨论质量评价中以检验模型的可行性和有效性。应用结果表明：在个体层面，使用逐步多元回归分析可建立评价维度与课程成绩的关系模型；在小组层面，使用聚类分析可挖掘小组协作模式及其动态迁移特点，同时依据评价指标可进一步识别出讨论过程中的意见领袖和边缘参与者行为；在班级层面，使用可视化分析和高频关键词分析可对比分析班级间的整体表现与差异。

Synchronous online discussion is an important means to promote learners' knowledge construction in online collaborative learning. The evaluation of synchronous online discussion quality helps learners and teachers to identify learning problems in a timely manner and make interventions and adjustments. However, current research on the evaluation of online discussion quality mostly focuses on asynchronous online discussion scenarios, and the evaluation dimensions, granularity and methods need to be expanded from a diversified perspective. Therefore, this paper proposes a quality evaluation framework for synchronous online discussion scenarios, which includes four evaluation dimensions of cognition, behavior, emotion, and social interaction, as well as three evaluation granularities of individual, group, and class. It also provides a method for automatically calculating and integrating evaluation indicators based on online discussion texts. To further leverage the learning-promoting role of evaluation, this paper constructs a multidimensional and multigranularity analysis model based on the evaluation framework, elaborates on the main process of evaluation work, and emphasizes the use of different learning analysis methods to analyze the characteristics and differences of individuals, groups, and classes in the four evaluation dimensions, achieving a deep evaluation of the quality of online discussions. Finally, to test the feasibility and effectiveness of the model, it was applied to the evaluation of synchronous online discussion quality in three classes. The application results show that at the individual level, a relationship model between evaluation dimensions and course grades can be established using stepwise multiple regression analysis; at the group level, clustering analysis can be used to explore group collaboration patterns and their dynamic transition characteristics, and the behavior of opinion leaders and marginal participants in the discussion process can be identified according to the evaluation index; at the class level, visualization analysis and high-frequency keyword analysis can be used to compare and analyze the overall performance and differences between classes.