Erik Sandewall – författare

Techniques for reasoning about actions an change in the physical world is one of the classical research topics in artificial intelligence. It is motivated by the needs of autonomous robots which must be able to anticipate their immediate future, to plan their future actions, and to figure out what went wrong in case of problems. It is also motivated by the needs of common-sense reasoning for example in the understanding of natural language texts, where processes and change over time is an ever-present phenomenon. The same set of problems arises in several other areas of computing such as in conceptual modelling for data bases, and in the rapidly growing area of intelligent control.The present research monograph presents and uses a novel methodology for reasoning about actions and change. Traditional research contributions have proposed new logic variants which were only supported by episodical examples. THe work described here uses a systematic methodology for identifying the exact range of applicability of a given logic. For a number of previously proposed logics, as well as for some new ones, the present work characterizes exactly the class where it does not.This book will be a necessary source of reference for researchers in knowledge representation, cognitive robotics, and intelligent control in the years to come. Particularly because of its emphasis on a strict and systematic methodology, it can also be recommended as a textbook for graduate university courses in these areas.

Principles of Knowledge Representation and Reasoning contains the proceedings of the Fourth International Conference on Principles of Knowledge Representation and Reasoning (KR ''94) held in Bonn, Germany, on May 24-27, 1994. The conference provided a forum for reviewing the theory and principles underlying knowledge representation and reasoning. Topics covered range from reasoning about mental states and spatial reasoning with propositional logics to default logic as a query language. Comprised of 60 chapters, this book begins with a description of a formal language for representing and reasoning about time and action before turning to proof in context and how it can replace the most common uses of reflection principles. The reader is then introduced to reasoning with minimal models; belief ascription and mental-level modeling; and a unified framework for class-based representation formalisms. A general approach to specificity in default reasoning is also described, together with an ontology for engineering mathematics and the use of abduction to generate tests. The book concludes by considering the use of natural language for knowledge representation and reasoning. This monograph will be of interest to both students and practitioners in the fields of artificial intelligence and computer science.