Case-based approximate reasonning. (Q859711)

Many years the developments in Artificial Intelligence have been based on logics (propositional and predicate calculus) and search methods (depth-first, breadth-first, heuristic search). Together with the always growing speed of the processors and the larger memory spaces available this development was very successful (as can be seen very well in Computer Chess and similar areas). Developments such as default reasoning and fuzzy logics tried to bridge the gap to real-world problems with incomplete information, the handling of inconsistencies, the formalization of approximations and similarity. In the last years developments were very successful that have been based on the general concept of case-based reasoning. This is a very logical step and very much oriented towards the solution of real-world problems, for very different reasons. First of all, for some of the problems it is nearly impossible to create formal models, because of the existing knowledge in the field or because of the complexity of the models to be created. On the other side in many areas of human activities there exist a huge number of examples (i.e. knowledge) that are case-based, for instance, in finance, in many areas of medicine, in legal systems of many countries where judgments are case-based, in games such as chess, go and many more. This knowledge is hardly utilized right now and can be used for decision-making. The situation is very often so that some knowledge about a situation and the diagnosis are exactly known, and human beings use the similarity to previous cases for their decision-making. Hence, this kind of ``intelligent behaviour'' is very human. The concept of \textit{similarity} is the basic concept of this approach, and it has to be defined for each application in such a way that it corresponds appropriately to the problem to be solved. Many ad hoc systems have been created, but according to my knowledge this book is the first book that tries to summarize many approaches and create a general theory of case-based reasoning. Therefore this book will get a lot of attention and for a good while will be the reference for many applications and further research. Step by step the following areas are combined, explored and presented: (1) similarity and case-based inference, (2) constraint-based modeling of case-based inference, (3) probabilistic modeling of case-based inference, (4) fuzzy set-based modeling of case-based inference, (5) case-based decision making. The concepts used are carefully explained, and the book can be used as an excellent guideline for the implementation of problem-solving programs, but also for courses in Artificial and Computional Intelligence. Everybody who is involved in research, development and teaching in Artificial Intelligence will get something out of it.