The aim of this paper is to present an empirical study for defining the appropriate triggers for personality related stereotypes. These stereotypes are used for modelling the students in a Computer Supported Collaborative Learning (CSCL) system for UML. The system builds individual student models to provide adaptive and intelligent advice concerning their knowledge and the most adequate colleagues for collaboration. Most of existing CSCL systems and Intelligent Learning Environments (ILE’s) also include student models. However, the vast majority of them use the student models to describe the students from the perspectives of knowledge and/or participation in collaborative activities. In our approach, the student models additionally to the knowledge describe the students regarding specific personality characteristics related to their learning and collaboration attitudes. The student models are built using the stereotype-based method, which entails the definition of the stereotypes, their facets and the triggers (the triggering conditions for a student to belong to a stereotype). As the definition of the triggers is a task of high importance for the effectiveness and accuracy of the student models, we conducted an empirical study among experienced trainers of software engineering.