An optimized group formation scheme to promote collaborative problem-based learning
An optimized group formation scheme to promote collaborative problem-based learning
复制标题
DOI：
10.1016/j.compedu.2019.01.011
复制
发表时间：
2019-05-01
影响因子：
12
通讯作者：
Kuo, Chi-Hsiung
中科院分区：
文献类型：
--
作者：
Chen, Chih-Ming;Kuo, Chi-Hsiung
小组形成是协作学习的关键过程之一，因为学习小组中有足够的成员可以支持成员之间良好的协作互动，并且是确保令人满意的学习绩效的基础。先前的几项研究提出了基于遗传算法的小组形成方案，该方案考虑多个学生特征来优化协作学习小组。然而，遗传算法（GA）中用于评估群体形成质量的适应度函数可能会确定具有不平衡学习特性的协作学习群体。此外，很少有研究考虑如何利用学习角色和同伴之间的互动来优化协作学习小组，并证实不同的小组形成方案对学习绩效和同伴互动的影响。因此，本文提出了一种基于遗传算法的带有惩罚函数的小组形成方案（GAGFS-PF），该方案考虑了学生知识水平和学习角色的异质性，以及学习小组成员之间通过社交网络分析测量的社交互动的同质性，以生成具有平衡学习特征的协作学习小组，以提高学生的学习成绩，并促进学生在基于问题的协作学习（CPBL）环境中的互动。本工作采用准实验研究方法收集定量数据，评估三种小组形成方案——提出的GAGFS-PF、随机小组形成方案和自选小组形成方案——对CPBL环境中学习表现和交互效果的影响，并采用访谈来增强定性数据分析的结果。即本研究采用混合研究来检验研究结果。台湾新北市一所小学六年级三个班的 83 名学生被邀请参加实验。三个班级被随机分配到三个实验组，使用不同的分组方案，包括提出的GAGFS-PF、随机分组方案和以“全球变暖”为主题的CPBL活动的自选分组方案。结果表明，在四个 CPBL 阶段中，在“行动 2”学习阶段由两名教师评估的完整报告中，所提出的 GAGFS-PF 明显优于随机分组方案。分析结果还表明，使用社交网络测量进行评估时，所提出的用于群体形成的 GAGFS-PF 在同伴互动的效果方面显着优于随机和自选的群体形成方案。访谈结果还表明，所提出的 GAGFS-PF 在确定协作学习组方面具有优势。这项工作为提高协作学习绩效提供了一种新颖且有用的小组形成方案，也有助于呼吁该领域的未来研究。
Group formation is one of the key processes in collaborative learning because having adequate members in the learning groups supports good collaborative interactions among members and is fundamental to ensuring satisfactory learning performance. Several previous studies have proposed genetic algorithm-based group formation scheme that considers multiple student characteristics to optimize collaborative learning groups. However, the fitness function used in the genetic algorithm (GA) for assessing the quality of group formation may determine collaborative learning groups with unbalanced learning characteristics. Additionally, few studies considered how learning roles and interactions among peers can be used to optimize collaborative learning groups and confirmed the effects of different group formation schemes on learning performance and peer interaction. Therefore, this work proposes a novel genetic algorithm-based group formation scheme with penalty function (GAGFS-PF) that considers the heterogeneous of student's knowledge levels and learning roles, and the homogeneity of social interactions measured by social network analysis among the members in the learning group, to generate collaborative learning groups with balanced learning characteristics for improving student's learning performance and facilitate student's interactions in a collaborative problem-based learning (CPBL) environment. This work uses a quasi-experimental research method to collect quantitative data to assess the effects of three group formation schemes - the proposed GAGFS-PF, the random group formation scheme, and the self-selection group formation scheme - on the learning performance and effects of interaction in a CPBL environment and also adopts interview to enhance the results of qualitative data analysis. Namely, this study adopts a mixed study to examine the research findings. Eighty-three students from three Grade 6 classes at an elementary school in New Taipei City, Taiwan were invited to participate in the experiment. Three classes were randomly assigned to the three experimental groups that used different group formation schemes including the proposed GAGFS-PF, random group formation scheme, and self-selection group formation scheme for CPBL activities on the topic of "global warming". The results reveal that the proposed GAGFS-PF is significantly superior to the random group formation scheme in the score of a completed report assessed by two teachers during the "action 2" learning stage, among the four CPBL stages. Analytical results also show that the proposed GAGFS-PF for group formation is significantly superior to the random and self-selection group formation schemes in the effects of peer interaction, as assessed using social network measures. The interview results also support that the proposed GAGFS-PF provides benefits in determining collaborative learning groups. This work contributes a novel and useful group formation scheme for enhancing collaborative learning performance and also helps in calling for future research in this field as well.