Autonomous dynamic reconfiguration in multi-agent systems. Improving the quality and efficiency of collaborative problem solving (Q1850463)

This book deals with a new approach to collaborative problem solving. The main difficulties in this area are -- the necessary communication with other agents in the problem-solving environment, -- the lack of global knowledge that can contribute to the solution of the problem. The main idea of the presented approach is the autonomous and dynamic distribution of resources, based on the problem itself, by melting and splitting agents. The collaborative structure of the set of agents is continuously adapted to the progress of the problem solution; this process is called ``autonomous dynamic reconfiguration'' and detailed very carefully. The necessary steps to a practical application of this approach are elaborated in a very impressive way. The infrastructure of multi-agent systems, their use for the solution of constraint problems and the implementation of control strategies for the agent melting and splitting process are very clear and logical. One special property of this book and this project is the continuous and consistent use of one example, the scheduling of medical appointments. This problem is very typical for problems that can be fully based on constraints and that are of a very high discrete sensitivity. A small change in the conditions can completely change the solution options and the solution itself. The used formalism is not easy to read, it requires careful studies, sometimes it seems to be a bit overdone, but the background of a Ph. D. thesis can justify this. For the reviewer, it was very nice to see that PROLOG is still very alive and can be successfully used for solving difficult problems. A comprehensive list of references and the usual high quality of the series make, together with the contents, this book a must for all AI researchers in this area.