Goal model driven alternative selection: a quantitative approach

Abstract

Model driven engineering (MDE) techniques can be used in requirement engineering to derive an implementation out of system requirements, which could be extended to derive models in the solution space out of models in the problem space. Goal models are useful to deal with problem space modeling and support requirements analysis activities including alternative selection, a procedure that is performed to evaluate the feasibility and desirability of alternative strategies with respect to quality goals. The results of alternative requirements selection can be referred to derive the configuration of solution space models and accordingly the implementation of software, since requirements elements can be traced to architecture elements or architecture design issues. Most of the existing goal-oriented requirement engineering (GORE) frameworks conduct alternative selection based on qualitative goal models, which are too coarse-grained to differentiate alternatives. Several works offer quantitative analysis based on quantified goal models, but they did not provide guided methods to obtain the numbers in these models. In this paper, we extend general goal models by appending quantitative attributes and a modified AHP-based approach to the quantification of goal-related links. Based on this quantitative goal model, an algorithm is proposed to guide the procedure of goal violation detection and multi-criteria alternative selection. We have evaluated the proposed approach by comparing it with six related approaches, with the conclusion that our method makes improvements to support multi-criteria selection of requirements and design alternatives.