Applying genetic classifier systems for the analysis of activities in collaborative learning environments

Abstract

The analysis of activities in CSCL (Computer-Supported Collaborative Learning) environments can provide us with some interesting conclusions about collaborative learning processes themselves. Specifically, such an analysis can show the effectiveness of such processes and allow for the definition of intervention mechanisms which can motivate and engage the students in the learning activities. Until now, this analysis has focused on the collaboration process and the resulting product separately. We hypothesize that the use of Artificial Intelligence techniques can be useful for the production of a rule-based system that considers both the process and the product. Of the existing techniques, we propose the use of Genetic Classifier Systems (GCS) for their ability to evolve and adapt. The use of these rules allows for the identification and characterization of learning situations, in addition to the generation of feedback that can guide the students and the group towards a more effective learning experience. At the same time, the rule system can adapt to the new learning activities. The analysis method proposed in this article focuses on CSCL environments in which the collaboration takes the form of a conversation. We also present a tool that implements this approach and the results of its application to some learning activities, using an environment for collaborative learning of design. © Wiley Periodicals, Inc. Comput Appl Eng Educ 21: 704–716, 2013