We present the PA-Training Assistant, designed to be a pedagogical support to make students train their peer assessment capabilities, at their pace, when the teacher is not available. The Assistant is designed to provide the student with a solution to a task, get her/his assessment of the solution, and provide a review of such assessment, with feedback. We envision this Assistant as a help, allowing students to train on their own, in between actual peer assessment sessions. The assistant is a custom GPT machine, prepared under ChatGPT. It is implemented using prompt principles, and archive-based resources. At the current stage of the Assistant’s development, we wish to investigate on the Assistant’s ability to support the student autonomously, so here we present a preliminary, and somewhat limited, analysis of similarity between Teacher’s and Assistant’s reviews of the peer assessments produced by a sample of students. The results tell of a moderate semantic similarity, which in our view is encouraging, allowing to see a path for improvement, towards an effective use for students and teachers benefit.

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

Training for Peer-Assessment: An Approach through Conversational Agents

  • Gabriel Badea,
  • Elvira Popescu,
  • Marco Temperini

摘要

We present the PA-Training Assistant, designed to be a pedagogical support to make students train their peer assessment capabilities, at their pace, when the teacher is not available. The Assistant is designed to provide the student with a solution to a task, get her/his assessment of the solution, and provide a review of such assessment, with feedback. We envision this Assistant as a help, allowing students to train on their own, in between actual peer assessment sessions. The assistant is a custom GPT machine, prepared under ChatGPT. It is implemented using prompt principles, and archive-based resources. At the current stage of the Assistant’s development, we wish to investigate on the Assistant’s ability to support the student autonomously, so here we present a preliminary, and somewhat limited, analysis of similarity between Teacher’s and Assistant’s reviews of the peer assessments produced by a sample of students. The results tell of a moderate semantic similarity, which in our view is encouraging, allowing to see a path for improvement, towards an effective use for students and teachers benefit.