Summary

A new approach for conceptualizing and modeling multi-agent systems that consist of people, devices, and software agents.

Today, when computing is pervasive and deployed over a range of devices by a multiplicity of users, we need to develop computer software to interact with both the ever-increasing complexity of the technical world and the growing fluidity of social organizations. The Art of Agent-Oriented Modeling presents a new conceptual model for developing software systems that are open, intelligent, and adaptive. It describes an approach for modeling complex systems that consist of people, devices, and software agents in a changing environment (sometimes known as distributed sociotechnical systems). The authors take an agent-oriented view, as opposed to the more common object-oriented approach. Thinking in terms of agents (which they define as the human and man-made components of a system), they argue, can change the way people think of software and the tasks it can perform. The book offers an integrated and coherent set of concepts and models, presenting the models at three levels of abstraction corresponding to a motivation layer (where the purpose, goals, and requirements of the system are described), a design layer, and an implementation layer. It compares platforms by implementing the same models in four different languages; compares methodologies by using a common example; includes extensive case studies; and offers exercises suitable for either class use or independent study.

Choice Review

The fast pace and increasing complexity of the technological age demand that computer scientists develop new ways of creating software. The concept of "software agents," as small intelligent entities to help create this software, has entered the software engineering process. Sterling (Univ. of Melbourne, Australia; editor, The Practice of Prolog, CH, Sep'91, 29-0358) and Taveter (Tallinn Univ. of Technology, Estonia) have extended the definition of "agent" to include people, devices, software, networks, etc., to view the development from a systems perspective and visualize how all of it will fit together. Using plenty of examples and illustrations, this well-written book will help readers understand how to model such systems and software. Topics include agent concepts; various models; embedding quality in the system; different development platforms, methodologies, and programming languages; industry applications; home applications; and e-learning applications. Development platforms are compared with modeling examples in four different programming languages. One weakness is that only four of the ten chapters include exercises, requiring instructors to develop their own problems if they want to use this as a textbook. This work is a valuable reference for this important new agent modeling development technique. Good index, list of references, acronym list, and glossary. Summing Up: Highly recommended. Lower-division undergraduates through professionals/practitioners; two-year technical program students. H. J. Bender independent scholar,