AILP: Abductive Inductive Logic Programming

Abstract

Inductive Logic Programming (ILP) is often situated as a research area emerging at the intersection of Machine Learning and Logic Programming (LP). This paper makes the link more clear between ILP and LP, in particular, between ILP and Abductive Logic Programming (ALP), i.e., LP extended with abductive reasoning. We formulate a generic framework for handling incomplete knowledge. This framework can be instantiated both to ALP and ILP approaches. By doing so more light is shed on the relationship between abduction and induction. As an example we consider the abductive procedure SLDNFA, and modify it into an inductive procedure which we call SLDNFAI. Keywords: Inductive Logic Programming, Abductive Logic Programming, Incomplete Knowledge, Intensional Knowledge Base Updating, Theory Revision. 1 Introduction It is often argued that the use of - a subset of - first order logic as a representation language situates Inductive Logic Programming (ILP) at the intersection of Logic Programmin...