iStar4RationalAgents: Modeling Requirements of Multi-agent Systems with Rational Agents

摘要

Multi-agent systems (MAS) involve a wide variety of agents that interact with each other to achieve their goals. Usually, the agents in a MAS can be reactive or proactive, this choice defines the rationale of its elements. Rational Agents is the term used to mention a set of four kinds of reactive and proactive agents. Conceptual models which represent the rational agents' intentionality can be used to design and analyze MAS in a systematic and structured manner. Conceptual modelling can be used to uncover mistakes and gaps in reasoning that are missed or obscured via ad hoc evaluation. However, the modelling of MAS with different rational agents is a non-trivial task, due to the specificity of their domain concepts, also at requirements level. This paper presents an approach to model MAS with rational agents in requirements level using iStar. This is part of a Model-Driven Development approach which has been proposed to support the development of MAS with rational agents involving requirements, architecture, code and test. We extended iStar to support the modelling of main concepts of this domain in a systematic way based on a process to conduct iStar extensions. We modelled a MAS to validate and illustrate the usage of our extension and evaluate the results using a survey with experienced researchers/developers in MAS.