(DBWORLD) New Book Announcement

Juergen Dix (dix@mailhost.uni-koblenz.de)
Wed, 28 May 1997 12:25:54 -0500 (CDT)

Two new books of potential interest for people working in knowledge
representation and reasoning have appeared recently:

G. Brewka, J. Dix, and K. Konolige: Nonmonotonic Reasoning - An Overview
CSLI Lecture Notes No. 73, CSLI publications, Stanford, 1997
(http://www-csli.stanford.edu/publications/nonmonotonic.html)

G. Brewka (ed.), Principles of Knowledge Representation, CSLI publications,
Studies in Logic, Language and Information, Stanford, 1996
(http://www.dcs.warwick.ac.uk/~mdr/SiLLI/silli-brewka.html)

Both books thus cover many of the fundamental questions and problems
of knowledge representation. They should be useful for everyone who
wishes to catch up with ongoing research in this exciting field.

The first book gives an up-to-date survey of research in the area of
nonmonotoninc reasoning. It includes a concise description of the most
influential nonmonotonic logics (e.g. circumscription, autoepistemic
logic and default logic), a presentation of recent research in
abduction, as well as an overview of semantics for logic programs with
default negation. The primary goal of the volume is to make recent
results in the field more accessible. An extensive bibliography is
included.

Here is the table of contents:

1 Introduction 1
1.1 The Major Application Problems 1
1.2 How This Book Is Organized 5

2 Preference Logics 9
2.1 Closed-World Assumption 10
2.2 Circumscription 12
2.3 Preferred Models 20
2.4 Conclusion 21

3 Nonmonotonic Inference Relations 23
3.1 Structural Properties and Cumulativity 24
3.2 Logical Connectives and Rationality 26
3.3 Metatheoretic Closures and Conditional Logics 31
3.4 Conclusion 36

4 Consistency-Based Logics 39
4.1 Default Logic 40
4.2 Modal Nonmonotonic Logics 51
4.3 Maximal Consistency Logics 54
4.4 Implementations 58
4.5 Conclusion 64

5 Abduction 65
5.1 Abduction in AI 65
5.2 Logic-Based Systems: Formulation 66
5.3 Assumption-Based Truth Maintenance 68
5.4 Abduction in Non-Horn Theories 71
5.5 Abduction, Minimization, and Default Logic 78
5.6 Conclusion 86

6 Semantics of Programs with Negation 87
6.1 Some Historical Remarks 88
6.2 Logic Programming Semantics 90
6.3 Nonmonotonic Reasoning Semantics 101
6.4 Conclusion 119

7 Nonmonotonicity in Logic Programming 121
7.1 Classical NML's versus NMR-Semantics 121
7.2 Classifying and Characterizing Semantics 125
7.3 Conclusion 142

Epilogue 143
Bibliography 145
Index 175

The second book is a collection of eight survey papers. It contains
the following contributions:

Didier Dubois and Henri Prade:
Non-Standard Theories of Uncertainty in Plausible Reasoning,

identifies several requirements reasoning with exceptions has to
satisfy. It is shown that neither classical logic nor Bayesian
networks satisfy all of these requirements. The authors then discuss a
number of non-standard approaches to uncertainty - both numerical and
non-numerical - in the light of these requirements. In particular,
they present possibilistic logic which also turns out to be a suitable
framework for encoding rational inference operations.

Moises Goldszmidt and Judea Pearl:
Probabilistic Foundations of Reasoning with Conditionals

Focuses on an approach which is based on infinitesimal conditional
probabilities and can be viewed as constituting a universal core for
reasoning with defaults. The authors show how the basic formalism can
be extended, using ranked models, maximum entropy and other
techniques, to obtain more ``adventurous'' conclusions. Moreover,
they demonstrate that this approach also helps to solve problems of
belief revision, causal reasoning, reasoning about action and
counterfactuals.

Vladimir Lifschitz:
Foundations of Logic Programming

presents the theory of ``extended'' logic programs, i.e., programs
containing both classical negation and negation as failure. Such
programs have proven to be of great importance for knowledge
representation. The focus of Lifschitz's paper is on the declarative
semantics of logic programming which is defined based on the notion of
an answer set.

Kurt Konolige:
Abductive Theories in Artificial Intelligence

investigates reasoning to explanations. After a logical analysis of
abduction, proof methods based on assumption based truth maintenance
and on resolution are discussed. Konolige then shows how abductive
methods can be used to implement special cases of circumscription and
default logic, and to generate causal explanations.

Stefan Wrobel:
Inductive Logic Programming

Whereas abduction generates an explanation of a particular observed
fact, induction tries to capture regularities in a certain set of data
and to produce a hypothesis (most often a general rule) that not only
explains what is known but also makes new predictions possible. Recent
work in machine learning has investigated induction in the setting of
first order logic and is surveyed in this paper. The paper also
presents some of the most important algorithms.

F.M. Donini, M. Lenzerini, D. Nardi and A Schaerf:
Reasoning in Description Logics

>From a semantical point of view description logics can be seen as
special cases of first order logic with a non-standard syntax that is
particularly well-suited for describing concept hierarchies. As it
turns out, the new syntax is not only convenient, it also led to new
inference methods and important complexity results. The paper surveys
these techniques and results in different settings: reasoning with
concepts; reasoning with concept instances; reasoning with axioms
describing concept properties; reasoning with axioms and concept
instances.

Bernhard Nebel:
Artificial Intelligence: A Computational Perspective

The role of computational complexity analysis for AI in general is the
topic of the paper. A large number of examples from areas like
temporal reasoning, planning, description logics, inheritance, theory
revision, and natural language processing are given. These examples
illustrate why complexity analysis can be extremely fruitful for AI.

Oskar Dressler and Peter Struss:
The Consistency-Based Approach to Automated Diagnosis of Devices

surveys one of the most promising application areas for knowledge
representation techniques. Theories that provide a principled approach
to diagnosing artifacts are presented together with several systems
that implement these approaches. The key idea underlying these systems
is to explicitly model the device structure and the behavior of its
constituents and to organize diagnosis as an inference process based
on this model and the observed behavior.

-------------------------------------------------------------------------------
The dbworld alias reaches many people, and should only be used for
messages of general interest to the database community.

Requests to get on or off dbworld should go to listproc@cs.wisc.edu.

to subscribe send
subscribe dbworld Your Full Name

to unsubscribe send
unsubscribe dbworld

to change your address
send an unsubscribe request from the old address
send a subscribe request from the new address

to find out more options send
help
------------------------------------------------------------------------FOOTER-