Technology seems to evolve “bottom up,” from low levels of abstraction to higher levels, for example, from well-developed methods and tools for implementation, to usable methods and tools for design, to potential methods and tools for requirements. This paper focuses on early requirements, which can be very abstract, intentional, and qualitative. It summarizes and updates the authors’ earlier work, includes a survey of related literature, describes a tool-supported framework for modeling early requirements, and presents a methodology for goal-model analysis that is interactive, iterative, and supports stakeholder involvement.

The detailed and thorough paper integrates existing work on goal-oriented models, presents a methodology for model analysis, and describes supporting tools. The authors define a formal model consisting of actors and intentions and several types of relationships. Labels (for satisfiability) apply to an intention and are propagated to related intentions by predicate calculus rules that express the semantics of the relationships. Rules are formulated for forward propagation, which corresponds to composition processes, and the harder backward propagation, which corresponds to decomposition processes. The former are used for questions such as, “Will a selection of alternatives enable the satisfaction of a goal?”; the latter are used for questions such as, “Is a goal satisfiable? By which alternatives?”

A methodology guides elicitation of intentions, creation of an instantiated goal model, and its analysis. It is supported by an open-source software framework and an off-the-shelf Boolean satisfiability (SAT) solver. An example application is used throughout the paper to explain and illustrate the application of the methodology. Lastly, evaluation and performance of model analysis and its implementation are discussed, including complexity, tests, and case studies.

Clearly, the authors’ research is challenging. It is notable for its early requirements focus, stakeholder involvement, iterative model improvement, tool support, leverage of related work, and the integration of paradigms (relational, logic, object-oriented, declarative). Many challenges remain: capture of the “right” stakeholder intentions using the “right” abstractions and interconnections, usability of “big” models, incorporation of cost/time/quality for comparison of alternatives, and interface of a goal model with late requirements and design models. Moreover, the scope of applicability is wide: business, medical, economic, societal, and so on. It may take many years for early requirements technology to mature; however, current research on goal models is encouraging. Perhaps technological evolution is moving from automation of the “how” and “what” to the “why,” bottom up.