Bücher der Reihe Artificial Intelligence

Natural language generation is a field within artificial intelligence which looks ahead to the future when machines will communicate complex thoughts to their human users in a natural way.

Logic grammars have found wide application both in natural language processing and in formal applications such as compiler writing. Major sections of the book include: grammars for formal language and linguistic research, writing a simple logic grammar, different types of logic grammars, applications, and logic grammars and concurrency.

The ideas and processes presented in this volume further our understanding of computational models of design, particularly those that are capable of assisting in the production of non-routine designs, and affirm that we are indeed moving toward a science of design.

With contributions by numerous experts

Knowledge representation is perhaps the most central problem confronting artificial intelligence. The problem of knowledge representation breaks down into several subsidiary problems including what knowledge to represent in a particular application.

In this paper, a comparison is made of several proof calculi in terms of the lengths of shortest proofs for some given formula of first order predicate logic with function symbols. In particular, we address the question whether, given two calculi, any derivation in one of them can be simulated in the other in polynomial time. The analogous question for propositional logic has been intensively studied by various authors because of its implications for complexity theory. And it seems there has not been as much endeavour in this field in first order logic as there has been in propositional logic. On the other hand, fOr most of the practical applications of logic, a powerful tool such as the language of first order logic is needed. The main interest of this investigation lies in the calculi most frequently used in automated theorem proving, the resolution calculus, and analytic calculi such as the tableau calculus and the connection method. In automated theorem proving there are two important aspects of complexity. In order to have a good theorem proving system, we must first have some calculus in which we can express our derivations in concise form. And second, there must be an efficient search strategy. This book deals mainly with the first aspect which is a necessary condition for the second since the length of a shortest proof always also gives a lower bound to the complexity of any strategy.

Indeed, the basic description of a plan generation algorithm has remained constant for nearly three decades: given a desciption of an initial state I, a goal state G, and a set of action types, find a sequence S of instantiated actions such that when S is executed instate I, G is guaranteed as a result.

Originally presented as the author's thesis (doctoral--Technical University, Munich, 1990).

They include explicit picture query languages that augment alphanumeric data query langua ges as well as languages and command sets which are implicitly embedded in a pictorial information system but perform similar functions.

Like a finite state automaton, an ATN consists of a collection of la beled states and arcs, a distinguished start state and a set of distin guished final states.

Gelernter finished a computer program to prove simple high school geometry theorems. The article by Martin Davis in the first of this series of volumes traces the most influential ideas back to the 'prehistory' of early logical thought showing how these ideas influenced the underlying concepts of most early automatic theorem proving programs.

This would seem to indicate that many of the essential aspects of diagnostic reasoning are a type of intuiti- based, common sense reasoning. More precisely, diagnostic reasoning can be classified as a type of inf- ence known as abductive reasoning or abduction.

Beyond this theoretical basis the scope of the book includes also implementational aspects and a valuation of existing models and systems.

A monograph for specialists interested in building maintainable knowledge based systems, giving a unified methodology for the design of such systems

It is a pleasure and an honor to be able to present this collection of papers to Ray Reiter on the occasion of his 60th birthday.

They address such diverse topics as: computational molecular biology, machine learning, mobile computing, multi-agent systems, planning, numerical computing and dynamical systems, database systems, an alternative to the "formulas as types" approach, program semantics and analysis, and natural language processing.

I am therefore delighted to see that machines of the power of the IBM PC can now run this system, even though these computers are more modestly priced than my 1980 vision of the personal algebra machine.

The purpose of this catalogue is to promote interaction between members of the AI' community. with no extended discussion of the historical origin of the tool or how it has been used in particular AI programs, The focus is on techniques abstracted from their historical origins.

Since the start of the DENDRAL project at Stanford University over 15 years ago, with its objective of problem-solving via the automation of actual human expert knowledge, significant expert systems projects have been completed in countries rang ing from Japan to France, Spain to China.

Finally, we have described in detail the extensions needed to the basic influence resolution algorithm in QP theory to oper ate on these extended descriptions. In the next chapter we extend these techniques further, by developing a qualitative perturbation analysis technique that permits us to answer "what ir' control questions;

Although the chess program was not complete even after ten years, the program PIONEER-2 for computing the monthly repair schedule for the Interconnected Power System of Russian Central was completed in a few months.

This is the first book that integrates nonmonotonic reasoning and belief change into a single framework from an artificial intelligence logic point-of-view. The approach to both these subjects is based on a powerful notion of an epistemic state that subsumes both existing models for nonmonotonic inference and current models for belief change.

Knowledge representation is perhaps the most central problem confronting artificial intelligence. The problem of knowledge representation breaks down into several subsidiary problems including what knowledge to represent in a particular application.

It also relates them to various modal logics studied in the philosophical logic litera ture, and provides a thorough treatment of their applications as foundations for logic programming semantics and for truth maintenance systems.