Empirically Driven Use Case Metamodel Evolution

Abstract

Metamodel evolution is rarely driven by empirical evidences of metamodel drawbacks. In this paper, the evolution of the use case metamodel used by the publicly available requirements management tool REM is presented. This evolution has been driven by the analysis of empirical data obtained during the assessment of several metrics–based verification heuristics for use cases developed by some of the authors and previously presented in other international fora. The empirical analysis has made evident that some common defects found in use cases developed by software engineering students were caused not only by their lack of experience but also by the expressive limitations imposed by the underlying use case metamodel used in REM. Once these limitations were clearly identified, a number of evolutionary changes were proposed to the REM use case metamodel in order to increase use case quality, i.e. to avoid those situations in which the metamodel were the cause of defects in use case specifications.