Improving the representation of roles in conceptual modeling: theory, method, and evidence

Conceptual models represent the Organizational domain for which an information system is developed. These models are important tools in defining the requirements for the system. When describing an Organization or part of it, a key concept is the notion of roles played by actors in the domain. Actors in an Organization act in various roles, hence, showing that roles in a conceptual model can promote understanding of how the Organization works. However, despite the importance of roles in understanding Organizations and their prevalence in various aspects of information systems development, no consensus exists on what roles are, or how to represent them in conceptual models. In this paper, we formally define role as a conceptual modeling construct based on literature analysis, ontological concepts, and principles of classification. Using this definition, we derive guidelines for representing roles in conceptual models and suggest rules for modeling roles with the widely used extended entity-relationship grammar. Finally, we test the effectiveness of the modeling rules by conducting an experimental study to compare the domain understanding of readers using two types of conceptual modeling scripts. One script was obtained by violating the rules and the other by not violating the rules. We obtained data on domain understanding (using problem-solving questions) and on the process of understanding (using eye tracking). The results indicate that the role-based rules are not only useful for understanding the models but also provide direct clues as to why this is so.