TimedGRL: Specifying Goal Models Over Time

Abstract

This thesis introduces TimedGRL, an extension of the current GRL (Goal-oriented Requirement Language) standard, which enables the modeling and analysis of a comprehensive set of changes to a goal model over time. Goal-oriented analysis, in general, deals with finding a solution to the modeled problem, which tries to accomplish desired system qualities while taking care of stakeholder needs simultaneously. It can be used to influence design decisions as it allows performing trade-off analysis among alternative solutions by determining which proposed solution strikes a more appropriate balance of stakeholder objectives and desired qualities of a system. Some common goal-oriented languages such as i*, GRL, the NFR Framework, and KAOS offer standard language definitions to support varying forms of such analysis. However, in most of these cases, the analysis is focused on only one snapshot in time. To capture evolving systems, several goal models need to be created that represent different stages of that system. The current language definitions do not allow these models to be connected and make it difficult to analyze them simultaneously. Moreover, it requires extra effort to maintain and consistently update these models that actually represent the same system and have many repeated elements.TimedGRL proposes capturing the goal model and the changes to the goal model in a single model to facilitate system evolution. The metamodel for TimedGRL is presented, and an analysis algorithm for TimedGRL models is defined and implemented in jUCMNav, the most comprehensive GRL modeling tool to date. A hypothetical but realistic example from the sustainability domain is used to illustrate the introduced concepts. Moreover, a detailed case study has been performed to show TimedGRL's functionality and usefulness.iii