Reviews - What do customers think about Understanding Agent Systems?
The Must-have Book about Agent Systems Nov 16, 2001
If you only buy one book on agent systems you could do a whole lot worse than splash out on this thoroughly admirable and admirably thorough work from d'Inverno and Luck. Easy to follow (even if your Z spec is a bit rusty, as mine is), clear and really well set out. The authors have a deep understanding of their subject and a rare ability to communicate their ideas.
Useful framework for agents, with helpful case studies. Aug 18, 2001
There are several books on intelligent agents and multi-agents systems that I've come across, but most are either too broad-ranging and shallow so that they don't actually get to important core issues, or they're too narrow and mathematical for my liking (and for many others). This excellent book somehow manages to pull off the feat of providing a good introduction to agents, while also drilling down to some fascinating and deep issues in multi-agent systems. What's particularly good is that it does two things - it analyses and explains the issues with really clear textual description, and then provides a more formal description (using the Z specification language) that is surprisingly readable.
After providing an introductory chapter, the book presents a "framework" for understanding agent systems (hence the title) in which it brings together various different notions of agents. The chapters cover the framework itself, the different kinds of inter-agent
relationships that arise within it (to get to multi-agent systems), and more complex agents with greater sophistication. There are also a couple of case-study chapters that show how the model can be used to give descriptions of BDI systems and the contract net.
Throughout, the authors provide really good explanations, and then also formal descriptions using Z. Whether or not you buy the claim that Z is the most used industrial formal method, it turns out that despite the mathematical nature of the Z specification, the book as a whole is really very readable. It is worth noting that the level of mathematical description in the book for describing the framework and the systems is pretty close to abstract code descriptions (which is perhaps not surprising given that Z is intended for use for specifying software). With the appendix intro to Z, the book should also be a useful resource for developers wanting to understand exactly what would be involved in building systems.
One of the difficulties I've found when reading about agents is trying to make sense of some very different ideas and systems, and trying to understand how they fit together. This book provides some of the answers. In summary, the book covers some basic agent concepts, and builds them up to describe quite complex multi-agent systems, moving from abstract ideas to descriptions of specific implemented systems, and showing how they come together. It provides an excellent
introduction to agents, and keeps going to address some much deeper issues.