An empirical assessment of using stereotypes to improve reading techniques in software inspections

Abstract

Stereotypes were introduced into the Unified Modeling Language (UML) in order to provide a means of customizing the language for particular needs. The stereotypes can increase the comprehension of UML diagrams and therefor influence reading techniques used for inspections of software artefacts. In this paper we evaluate how the usage of stereotypes in UML designs influences outcomes of three reading techniques used for verification and validation of UML models. The study presented in this paper is done in the context of the UML domain modeling, but the results can be generalized to other kinds of models expressed in UML. The results show that the presence of stereotypes improves the efficiency and effectiveness of the studied methods and shows the magnitude of the improvements. We also investigate which of the reading techniques are the most efficient and effective for analysis of UML designs with stereotypes.