PRISE : a process to support iStar extensions

Abstract

iStar is a goal-oriented requirements modelling language which has been used by industrial and academic projects of different domains. Modelling languages are commonly extended to add new constructs, which better represent specific concepts, giving more expressiveness to the extended language. iStar is often extended to incorporate new constructs from specific application areas or to adjust them to practical situations during requirements modelling. A study performed on iStar extensions identified the occurrence of problems related to their quality, such as the incompleteness related to absence of definition of meaning of the concepts introduced and metamodel; inconsistencies between extension representation levels; and conflicts between the graphical representations of the constructs. The iStar syntax has been recently standardised. However, the language continues to be extended. Thus, this is an appropriate time to propose a way to support the proposal of iStar extensions systematically to prevent the problem occurrence, increase the quality of extensions, and make extension creation a less challenging task. This thesis investigates how iStar extensions have been created and proposes a systematic way to guide the creation of quality extensions and avoid the occurrence of common problems. A set of empirical studies was carried out in this work. The identification of existing extensions was performed through a systematic review of the literature, which originated a catalogue of extensions of iStar that can facilitate the identification and reuse of extensions and constructs. A set of guidelines on how to create good extensions was established based on the results of an interview-based study and survey with researchers who have already extended iStar. Besides, a comprehensive literature review was also conducted to provide a theoretical framework. Finally, PRISE (PRocess to support IStar Extensions) was proposed, based on the background of modelling languages and the results of the empirical studies. It establishes a sequence of tasks and artefacts to be used to guide the creation of iStar extensions. One tool has been proposed to support the management of new extensions based on PRISE. The process was illustrated by creating one new extension. It was evaluated through interviews and a questionnaire with researchers with experience in creating iStar extensions and through a case study with a novice extender. PRISE has proven to be adequate to avoid the problems identified in previous extensions, offering a step-by-step approach to creating new extensions or fixing problems in existing ones.