How do agents affect modifiability? A comparison between two architectures for intelligent virtual environments for training

Abstract

The use of agents is spreading as a means to develop different kinds of software systems, among which we can find Intelligent Virtual Environments for Training. The agent community has already started to pay attention to software engineering issues to develop agent-oriented systems, but they are mainly focused on methodologies and, to some extent, design patterns. However, not much attention has been paid to software architecture for the moment. We compare two agent-based software architectures for Intelligent Virtual Environments for Training that are intended to be easily extended and modified. The first one was designed using an organizational approach recommended by some agent oriented methodologies. The second one is a redesign of the first architecture using more formal principles and methods of software architecture design. A comparison between both architectures highlights the need to pay more attention to software architecture design in this field. © 2008 Springer-Verlag Berlin Heidelberg.