An Automatic Identification of Team Roles in Forums

The success of a collaborative learning experience and the coordination of the work depend on appropiate performance of team roles. This paper introduces an intelligent system able of recognizing team roles manifested by students who interact in a virtual forum. The system has two kind of software agents: a group agent and personal agents. Nine Bayesian networks were specially developed to detect automatically the team roles manifested during the collaboration dynamic. The system was validated through collaborative experiences in forums with university students. The stored interactions were processed and the roles played by each group member were identified. In order to evaluate the effectiveness in the automatic recognition of team roles, the roles identified by Bayesian networks were compared with roles allocated manually by an educational psychologist. This intelligent system will faciliate to teachers the monitoring of work dynamics in forums.