Inductive Logic Programming is a young and rapidly growing field combining machine learning and logic programming. This self-contained tutorial is the first theoretical introduction to ILP; it provides the reader with a rigorous and sufficiently broad basis for future research in the area.
In the first part, a thorough treatment of first-order logic, resolution-based theorem proving, and logic programming is given. The second part introduces the main concepts of ILP and systematically develops the most important results on model inference, inverse resolution, unfolding, refinement operators, least generalizations, and ways to deal with background knowledge. Furthermore, the authors give an overview of PAC learning results in ILP and of some of the most relevant implemented systems.
Author(s): Shan-Hwei Nienhuys-Cheng, Roland de Wolf (auth.)
Series: Lecture Notes in Computer Science 1228 : Lecture Notes in Artificial Intelligence
Edition: 1
Publisher: Springer-Verlag Berlin Heidelberg
Year: 1997
Language: English
Pages: 410
Tags: Artificial Intelligence (incl. Robotics); Mathematical Logic and Formal Languages; Programming Techniques
Propositional logic....Pages 2-15
First-order logic....Pages 17-34
Normal forms and Herbrand models....Pages 35-53
Resolution....Pages 55-74
Subsumption theorem and refutation completeness....Pages 75-92
Linear and input resolution....Pages 93-103
SLD-resolution....Pages 105-126
SLDNF-resolution....Pages 127-159
What is inductive logic programming?....Pages 162-177
The framework for model inference....Pages 179-195
Inverse resolution....Pages 197-206
Unfolding....Pages 207-217
The lattice and cover structure of atoms....Pages 219-242
The subsumption order....Pages 243-263
The implication order....Pages 265-278
Background knowledge....Pages 279-297
Refinement operators....Pages 299-320
PAC learning....Pages 321-343
Further topics....Pages 345-363
