Reusability in goal modeling: A systematic literature review

Abstract Context: Goal modeling is an important instrument for the elicitation, specification, analysis, and validation of early requirements. Goal models capture hierarchical representations of stakeholder objectives, requirements, possible solutions, and their relationships to help requirements engineers understand stakeholder goals and explore solutions based on their impact on these goals. To reuse a goal model and benefit from the strengths of goal modeling, we argue that it is necessary (i) to make sure that analysis and validation of goal models is possible through reuse hierarchies, (ii) to provide the means to delay decision making to a later point in the reuse hierarchy, (iii) to take constraints imposed by other modeling notations into account during analysis, (iv) to allow context dependent information to be modeled so that the goal model can be used in various reuse contexts, and (v) to provide an interface for reuse. Objective: In this two-part systematic literature review, we (i) evaluate how well existing goal modeling approaches support reusability with our five desired characteristics of contextual and reusable goal models, (ii) categorize these approaches based on language constructs for context modeling and connection to other modeling formalisms, and then (iii) draw our conclusions on future research themes. Method: Following guidelines by Kitchenham, the review is conducted on seven major academic search engines. Research questions, inclusion criteria, and categorization criteria are specified, and threats to validity are discussed. A final list of 146 publications and 34 comparisons/assessments of goal modeling approaches is discussed in more detail. Results: Five major research themes are derived to realize reusable goal models with context dependent information. Conclusion: The results indicate that existing goal modeling approaches do not fully address the required capabilities for reusability in different contexts and that further research is needed to fill this gap in the landscape of goal modeling approaches.