Scholarlink

<p>To promote deeper cognitive interactions and positive socio-emotional interactions among group members, thereby achieving high-quality collaborative outcomes, researchers have endeavored to develop conversational agents (CAs) that provide adaptive support to small groups. However, existing CAs for supporting collaborative learning struggled to integrate and analyze multi-source learning data throughout the collaborative learning process and to offer comprehensive and personalized scaffolding based on diagnostic results. To address these issues, this study designed a CA named CollaBot, based on the contingent teaching model, which has been used in the past to guide teachers on how to provide adaptive scaffolding to small groups. CollaBot integrates AI technologies, including retrieval-based models, generative AI models, and retrieval-augmented generation techniques, to offer adaptive cognitive, metacognitive, and social scaffolding to groups engaged in online collaborative learning. A randomized controlled experimental design was employed, recruiting 78 undergraduate students who were randomly divided into two groups: the experimental group (n = 39) utilized a co-writing platform with CollaBot, while the control group (n = 39) used the platform with task scripts. Results indicate that students supported by CollaBot demonstrated significantly better learning performance. In addition, both CollaBot and task scripts significantly enhanced students’ self-efficacy for writing. Furthermore, analysis of the interview data revealed both positive perceptions about CollaBot, such as aiding group members in regulating their own and the group’s learning processes and supporting the development of writing skills, as well as negative perceptions, including causing anxiety and providing ambiguous feedback. This research provides guidance for the design of CAs and offers insights into harnessing hybrid intelligence between teachers and GAI to support collaborative learning.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A conversational agent based on contingent teaching model to support collaborative learning activities: impacts on students’ learning performance, self-efficacy and perceptions

  • 期刊论文

  • 发表日期：

    2025年7月14日

查看全文
  • Wanqing Hu,
  • Rushi Gong,
  • Sisi Wu,
  • Yanyan Li

摘要

To promote deeper cognitive interactions and positive socio-emotional interactions among group members, thereby achieving high-quality collaborative outcomes, researchers have endeavored to develop conversational agents (CAs) that provide adaptive support to small groups. However, existing CAs for supporting collaborative learning struggled to integrate and analyze multi-source learning data throughout the collaborative learning process and to offer comprehensive and personalized scaffolding based on diagnostic results. To address these issues, this study designed a CA named CollaBot, based on the contingent teaching model, which has been used in the past to guide teachers on how to provide adaptive scaffolding to small groups. CollaBot integrates AI technologies, including retrieval-based models, generative AI models, and retrieval-augmented generation techniques, to offer adaptive cognitive, metacognitive, and social scaffolding to groups engaged in online collaborative learning. A randomized controlled experimental design was employed, recruiting 78 undergraduate students who were randomly divided into two groups: the experimental group (n = 39) utilized a co-writing platform with CollaBot, while the control group (n = 39) used the platform with task scripts. Results indicate that students supported by CollaBot demonstrated significantly better learning performance. In addition, both CollaBot and task scripts significantly enhanced students’ self-efficacy for writing. Furthermore, analysis of the interview data revealed both positive perceptions about CollaBot, such as aiding group members in regulating their own and the group’s learning processes and supporting the development of writing skills, as well as negative perceptions, including causing anxiety and providing ambiguous feedback. This research provides guidance for the design of CAs and offers insights into harnessing hybrid intelligence between teachers and GAI to support collaborative learning.