Multiagent Systems: A Modern Approach to Distributed Artificial Intelligence

This is one out of very few books that exist on multiagent systems and distributed artificial intelligence which won't take for granted the reader. It has really taken me from elementary concepts up to industrial applications, passing over the actual "nitty gritty" of programming an agent. Many papers on artificial intelligence are just too mathematical for many readers, and those which have any application would describe what the program does, but only sometimes you get into knowing *how* the program does the trick. I personally would say the book is a "must-have-it", not only for those seeking to understand in concrete distributed artificial intelligence or multiagent systems, but also for any mathematician, physicists, computer scientists and engineer whose background touches somehow the concepts of artificial intelligence as a whole. Come to think of it, even any cunning philosopher of science could take real advantage of this book. Now then, concerning those industries which would like to have this book on their shelves, they would have to take in mind that this book is a departure point. A good one, but nevertheless only the start for deepening towards a concrete application the industry would have in mind.

Not a beginner's book: The technologies presented are relevant to (gasp) some real life problems. The treatment of Distributed Constraint Satisfaction problems was well written and usable. The initial treatment of all subjects was broad, delving into details after laying a sound foundation. Assumes a bit of mathematical sophistication on the part of the reader, but on the whole, well written, well organized and well worth it.

While perhaps this should not be your first book in AI, it should definitely be your first book on agents. It took me years to accumulate the knowledge present in this book, and it's a great survey of the field for the beginning investigator. However, while this is a great way to get started in agents, understand how agents can be used in intelligent applications (from distributed AI type problem solving, to resource optimization problems), and "get the agent paradigm," coverage of agent software engineering techniques is light. This is not a fault of this book - you will simply need other resources to understand the difficulties and current research in engineering multi agent systems (or indeed any concurrent distributed system). So if you are trying to figure out what to do with agents, or how agent systems work, this is the book for you. If you already know that and want help with formal specification of an agent, verifying your agents meet the specifications, etc. this is not the book for you. Understand that to build systems you will need both!

I have been using this book probably for an year now, and this has become indispensable to my work. The collection of authors in this book reads as an 'who's who' in multiagent systems and distributed AI research. Although, its' difficult to maintain continuity in edited books, the editor has done an commendable job. The authors provide a readable introduction to their area of expertise, and supplement them with an excellent bibliography...enough to get u started fast.

Thogh edited the book looks like a single authored text. The definitions are clearly presented with illustrations. The presentation is highly comprehensive. The book is thus highly readable with minimum effort. I liked the chapters on distributed problem solving and planning (Chapter 3),Learning in multiagent systems (chapter 6), Formal methods in DAI: Logic based representation and reasoning (chapter 8)and Groupware and computer supported cooperative work (chapter 10) very much.Readers of diverse interest in distributed artificial intelligence will find the other chapters equally interesting and useful.The book should be on the desk of anyone interested to learn the concepts of multiagent systems and technology.