FOIL is a first-order learning system that uses information in a collection of relations to construct theories expressed in a dialect of Prolog. This paper provides an overview of the principal ideas and methods used in the current version of the system, including two recent additions. We present examples of tasks tackled by FOIL and of systems that adapt and extend its approach. 1. Introduction All symbolic machine learning leads to the formulation or modification of theories, so the language in which theories are expressed is an important consideration. Firstorder theory languages have been used for at least thirty years, as documented by Sammut [1993]. Explanation-based generalisation systems [Mitchell, Keller and Kedar-Cabelli, 1986; DeJong and Mooney, 1986] have always required them, but the early and influential work of Shapiro [1983] and Sammut and Banerji [1986] also employed them in an inductive learning context. Nevertheless, first-order empirical learning, including...

Abstract not found

Contents Preface iii 1 What is Inductive Logic Programming? 1 1.1 The importance of learning : : : : : : : : : : : : : : : : : : : : : 1 1.2 Inductive learning : : : : : : : : : : : : : : : : : : : : : : : : : : 2 1.3 The problem setting for ILP : : : : : : : : : : : : : : : : : : : : : 4 1.4 Other problem settings : : : : : : : : : : : : : : : : : : : : : : : : 9 1.5 A brief history of the field : : : : : : : : : : : : : : : : : : : : : : 10 1.6 An outline of the thesis : : : : : : : : : : : : : : : : : : : : : : : 12 2 The Subsumption Theorem for Several Forms of Resolution 13 2.1 Introduction : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 13 2.2 Preliminaries : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 16 2.3 The Subsumption Theorem : : : : : : : : : : : : : : : : : : : : : 18 2.3

Inductive Logic Programming (ILP) can be viewed as research in the intersection of Logic Programming and inductive Machine Learning. Informally speaking the field is concerned with the induction of PROLOG programs. Being able to express the discovered knowledge in a first-order logic representation language can overcome some of the limitations of classical learning algorithms. The representational power of these algorithms is usually restricted to propositional domain theories such as decision trees in the well-known ID3 family (Quinlan 1986) or propositional Horn clauses as in AQ (Michalski, Mozetic, Hong, and Lavrac 1986) or CN2 (Clark and Niblett 1989). ILP algorithms, on the other hand, can not only test attributes for specific values, but also make use of relations (like equality) between the values of d...