The use of mathematical logic as a formalism for artificial intelligence was recognized by John McCarthy in 1959 in his paper on Programs with Common Sense. In a series of papers in the 1960's he expanded upon these ideas and continues to do so to this date. It is now 41 years since the idea of using a formal mechanism for AI arose. It is therefore appropriate to consider some of the research, applications and implementations that have resulted from this idea. In early 1995 John McCarthy suggested to me that we have a workshop on Logic-Based Artificial Intelligence (LBAI). In June 1999, the Workshop on Logic-Based Artificial Intelligence was held as a consequence of McCarthy's suggestion. The workshop came about with the support of Ephraim Glinert of the National Science Foundation (IIS-9S2013S), the American Association for Artificial Intelligence who provided support for graduate students to attend, and Joseph JaJa, Director of the University of Maryland Institute for Advanced Computer Studies who provided both manpower and financial support, and the Department of Computer Science. We are grateful for their support. This book consists of refereed papers based on presentations made at the Workshop. Not all of the Workshop participants were able to contribute papers for the book. The common theme of papers at the workshop and in this book is the use of logic as a formalism to solve problems in AI.

Principles for constructing intelligent systems Design of Logic-based Intelligent Systems develops principles and methods for constructing intelligent systems for complex tasks that are readily done by humans but are difficult for machines. Current Artificial Intelligence (AI) approaches rely on various constructs and methods (production rules, neural nets, support vector machines, fuzzy logic, Bayesian networks, etc.). In contrast, this book uses an extension of propositional logic that treats all aspects of intelligent systems in a unified and mathematically compatible manner. Topics include: * Levels of thinking and logic * Special cases: expert systems and intelligent agents * Formulating and solving logic systems * Reasoning under uncertainty * Learning logic formulas from data * Nonmonotonic and incomplete reasoning * Question-and-answer processes * Intelligent systems that construct intelligent systems Design of Logic-based Intelligent Systems is both a handbook for the AI practitioner and a textbook for advanced undergraduate and graduate courses on intelligent systems. Included are more than forty algorithms, and numerous examples and exercises. The purchaser of the book may obtain an accompanying software package (Leibniz System) free of charge via the internet at leibnizsystem.com.

Logic and its components (propositional, first-order, non-classical) play a key role in Computer Science and Artificial Intelligence. While a large amount of information exists scattered throughout various media (books, journal articles, webpages, etc.), the diffuse nature of these sources is problematic and logic as a topic benefits from a unified approach. Logic for Computer Science and Artificial Intelligence utilizes this format, surveying the tableaux, resolution, Davis and Putnam methods, logic programming, as well as for example unification and subsumption. For non-classical logics, the translation method is detailed. Logic for Computer Science and Artificial Intelligence is the classroom-tested result of several years of teaching at Grenoble INP (Ensimag). It is conceived to allow self-instruction for a beginner with basic knowledge in Mathematics and Computer Science, but is also highly suitable for use in traditional courses. The reader is guided by clearly motivated concepts, introductions, historical remarks, side notes concerning connections with other disciplines, and numerous exercises, complete with detailed solutions, The title provides the reader with the tools needed to arrive naturally at practical implementations of the concepts and techniques discussed, allowing for the design of algorithms to solve problems.

The use of mathematical logic as a formalism for artificial intelligence was recognized by John McCarthy in 1959 in his paper on Programs with Common Sense. In a series of papers in the 1960's he expanded upon these ideas and continues to do so to this date. It is now 41 years since the idea of using a formal mechanism for AI arose. It is therefore appropriate to consider some of the research, applications and implementations that have resulted from this idea. In early 1995 John McCarthy suggested to me that we have a workshop on Logic-Based Artificial Intelligence (LBAI). In June 1999, the Workshop on Logic-Based Artificial Intelligence was held as a consequence of McCarthy's suggestion. The workshop came about with the support of Ephraim Glinert of the National Science Foundation (IIS-9S2013S), the American Association for Artificial Intelligence who provided support for graduate students to attend, and Joseph JaJa, Director of the University of Maryland Institute for Advanced Computer Studies who provided both manpower and financial support, and the Department of Computer Science. We are grateful for their support. This book consists of refereed papers based on presentations made at the Workshop. Not all of the Workshop participants were able to contribute papers for the book. The common theme of papers at the workshop and in this book is the use of logic as a formalism to solve problems in AI.

Intended both as a text for advanced undergraduates and graduate students, and as a key reference work for AI researchers and developers, Logical Foundations of Artificial Intelligence is a lucid, rigorous, and comprehensive account of the fundamentals of artificial intelligence from the standpoint of logic. The first section of the book introduces the logicist approach to AI--discussing the representation of declarative knowledge and featuring an introduction to the process of conceptualization, the syntax and semantics of predicate calculus, and the basics of other declarative representations such as frames and semantic nets. This section also provides a simple but powerful inference procedure, resolution, and shows how it can be used in a reasoning system. The next several chapters discuss nonmonotonic reasoning, induction, and reasoning under uncertainty, broadening the logical approach to deal with the inadequacies of strict logical deduction. The third section introduces modal operators that facilitate representing and reasoning about knowledge. This section also develops the process of writing predicate calculus sentences to the metalevel--to permit sentences about sentences and about reasoning processes. The final three chapters discuss the representation of knowledge about states and actions, planning, and intelligent system architecture. End-of-chapter bibliographic and historical comments provide background and point to other works of interest and research. Each chapter also contains numerous student exercises (with solutions provided in an appendix) to reinforce concepts and challenge the learner. A bibliography and index complete this comprehensive work.

This book explores the building of expert systems using logic for knowledge representation and meta-level inference for control. It presents research done by members of the expert systems group of the Department of Artificial Intelligence in Edinburgh, often in collaboration with others, based on two hypotheses: that logic is a suitable knowledge representation language, and that an explicit representation of the control regime of the theorem prover has many advantages. The editors introduce these hypotheses and present the arguments in their favor They then describe Socrates' a tool for the construction of expert systems that is based on these assumptions. They devote the remaining chapters to the solution of problems that arise from the restrictions imposed by Socrates's representation language and from the system's inefficiency. The chapters dealing with the representation problem present a reified approach to temporal logic that makes it possible to use nonstandard logics without extending the system, and describe a general proof method for arbitrary modal logics. Those dealing with the efficiency problem discuss the technique of partial evaluation and its limitations, as well as another possible solution known as assertion-time inference. Peter Jackson is a Senior Scientist in the Department of Applied Mathematics and Computer Sciences at the McDonnell Douglas Research Laboratory in St. Louis. Han Reichgelt is a Lecturer in Department of Psychology at the University of Nottingham. Frank van Harmelen is a Research Fellow in the Mathematical Reasoning Group at the University of Edinburgh.

cians concerned with using logical tools in philosophy have been keenly aware of the limitations that arise from the original con centration of symbolic logic on the idiom of mathematics, and many of them have worked to create extensions of the received logical theories that would make them more generally applicable in philosophy. Carnap's Testability and Meaning, published in 1936 and 1937, was a good early example of this sort of research, motivated by the inadequacy of first-order formalizations of dis 'This sugar cube is soluble in water'. positional sentences like And in fact there is a continuous history of work on this topic, extending from Carnap's paper to Shoham's contribution to the present volume . . Much of the work in philosophical logic, and much of what has appeared in The Journal of Philosophical Logic, was mo tivated by similar considerations: work in modal logic (includ ing tense, deontic, and epistemic logic), intensional logics, non declaratives, presuppositions, and many other topics. In this sort of research, sin.ce the main point is to devise new formalisms, the technical development tends to be rather shallow in comparison with mathematical logic, though it is sel dom absent: theorems need to be proved in order to justify the formalisms, and sometimes these are nontrivial. On the other hand, much effort has to go into motivating a logical innovation.