Agent-Oriented Software Development Methodologies

  University of Toronto, CANADA   

 

    

    York University, Toronto, Canada                  

Quick Links 

List of Methodologies

The Exemplar

Despite rapid advances in agent technologies, their adoption in mainstream application areas such as large-scale information systems is still limited. It is generally recognized that a major reason is the lack of systematic methods to guide the development of agent-oriented systems.  Agent-oriented methodologies have thus become an important and urgent area of research. In the recent past, more than a dozen methodologies have been proposed.  They offer a range of modelling concepts, elaboration and analysis techniques, and opportunities for tool support. They vary in maturity and scope of coverage  The diversity of approaches offers rich resources for developers to draw on, but can also be a hindrance to progress if their commonalities and divergences are not readily understood.

One way to advance the state of research in agent-oriented methodologies is to define a suitable example problem that can serve as a focal point for discussion and exchange of research ideas and results.  Examples are indispensable for illustrating and explaining methodologies.  They provide concrete instances that allow abstract concepts and descriptions to come alive through domain settings and scenarios. Understandably, each methodology would typically be illustrated with examples that are best suited (possibly tailored) for demonstrating its specific features.  They are likely to emphasize certain aspects of the problem domain more than others. 

A common example that has become familiar to a research community can be a valuable resource in a number of ways.

bulletIt can be used to capture and embody the set of problems faced by the community.  It can present the problems with a clear separation from any solution approaches.  Specific research projects can use it to delineate research objectives without necessarily addressing the entire set of problems.
bulletIt can provide a familiar application context as a basis for defining and discussing the methodological issues.  Knowledge and experience gained about the context can be “reused,” allowing attention to focus on the solutions to problems.
bulletFor someone learning about a new methodology, the availability of a familiar example can help quickly place the methodology in relation to other methodologies, to understand what problems it addresses, and how it advances beyond existing ones.
bulletFor someone developing a methodology, it provides a more objective vehicle for assessing capabilities and limitations.  It can also be invaluable for third-parties to compare and evaluate different methodologies according to some criteria.
bulletIt can also be a great help to someone from outside the research community, to quickly grasp the issues faced, to see how well they are addressed by the state of the research. Potential users (methodology adopters) can use it to relate to their own domain problems, thus recognizing the strengths and limitations of agent-oriented approaches versus other approaches.

A common example can therefore be a useful instrument for a research community to help clarify research objectives, consolidate results, and focus future directions.

Common examples have been used to advantage in a number of areas in information systems and software engineering – in IS development methodologies, e.g., [Olle 82], in software specification, e.g., [Wing 88], in software architecture e.g., [Parnas 72], and in software process modelling [ISPW 93], in requirements engineering [Feather97].  [Feather97] has examined the use of such examples in detail, and refers to these as “exemplars”.  We will follow this usage.

We propose the use of a health care information systems example as an exemplar for the agent-oriented methodology community. The exemplar aims to be problem-centered, neutral with respect to particular methodologies or approaches, and rich in embodying real-world issues and challenges.  It is intended to test the limits of agent-oriented methodologies, allowing their strengths as well as weaknesses to come through. It is a challenge to individual methodologies to further their strengths and remedy their weaknesses.  It is a challenge to the community to combine the best ideas from various research results to synthesize practical methodologies for adoption by practitioners.

bulletThe Exemplar
bulletList of Methodologies
 

This version of the exemplar is based on the paper: "Agent-Oriented Methodologies – Towards A Challenge Exemplar" by Eric Yu    and   Luiz Marcio Cysneiros, to be presented at AOIS  workshop at CAiSE 2002, Toronto.

This page is maintained by:

Eric Yu  - Faculty of Information Studies

Luiz Marcio Cysneiros - School of Information Technology - Faculty of Atkinson - York University Canada

Please send inquires to Luiz Marcio

   UNDER CONSTRUCTION

Last Update: 15/10/2002