A Multidimensional Weighted-Attributes Framework (MWAF) for Evaluating Agent-Oriented Software Engineering Methodologies

Agent-oriented software engineering (AOSE) is concerned with the use of agents in the development of distributed systems, especially in open and dynamic environments that require autonomous processes with communicative and cooperative features. In this context, AOSE came to the forefront as a new discipline to assist the engineering of complex software systems based on the multi-agent systems paradigm. As a result, a large number of AOSE methodologies and modeling techniques evolved to support the development of agent-based applications in all of their life cycle phases. Deciding what agent-oriented software engineering (AOSE) methodology is the best to adopt for developing a potential multi-agent system depends on which aspects are considered more important for the prospective application. However, up till now, there is no industry-wide agreement on the kinds of concepts a methodology should support. This lack of standardization makes the research in this area an open issue with accompanying challenges. In this paper, we propose the multidimensional weighted-attributes framework (MWAF), a new framework that provides a reliable solution with accurate results based on applying state-of-the-art statistical procedures to evaluate AOSE methodologies and come up with a set of measures that help in selecting the most appropriate methodology for developing prospective agent-based applications