Multi agent model for skills training of CSCL e-tutors: modelo multi agente para el entrenamiento de habilidades de e-tutores de ACSC

Abstract

Computer Supported Collaborative Learning (CSCL) systems enable not only group learning with independence of the time and space where group members are located, but also they are favorable environments for leadership skills development. However, as interactions that are ideal for learning do not occur spontaneously, participation of e-tutors (teachers) is essential in order to generate interactions that contribute to collaborative building of knowledge. Considering e-tutors of CSCL usually do not know the most effective way to assist students, this article proposes a multi agent model (combining techniques from natural language processing, text mining, and machine learning) that can be used for personalized training of e-tutors. In the proposed model an intelligent agent analyzes group interactions to identify conflicts which resolution needs e-tutors' intervention. In these cases, a training agent suggests to e-tutors necessary actions so as they solve conflicts and simultaneously they develop skills they do not manifest properly. The multi agent model will be implemented in a CSCL environment and its operation will be evaluated through experiments with university students and teachers.