Connecting Methodologies and Infrastructures in the Development of Agent Systems

In the building of agent systems developers canbe supported by both appropriate methodologies and infrastructures, which guide them in the different phases of thedevelopment and provide useful abstractions.Nevertheless, we assist to a situation in which methodologiesand infrastructures are not connected each other: the products ofthe analysis and design phases could not always be exploited inthe implementation phase in a direct way, even if sometimesCASE-tools are present to help in translating methodologies’diagrams in infrastructures’ code. This leads to a “gap” betweenmethodologies and infrastructures that is likely to producefragmented solutions and to make the application maintenancedifficult.In this paper we face this issue, proposing three directions tosolve the problem. We do not want to propose a “new brand”methodology and infrastructure tightly connected, rather, weaim at reusing as much as possible what already exists, notonly in abstract terms, but also in concrete “fragments” ofmethodologies; an appropriate meta-language that describes howa methodologies works would be useful to more easily mapthem onto the infrastructures, or even to “compose” a newmethodologies. A further approach is based on an “intermediate”layer between methodologies and infrastructures, which providesa mapping between the involved entities.