Adopting an Object-Oriented Data Model in Inductive Logic Programming

The increasing amount of information to be managed in knowledge-based systems has promoted, on one hand, the exploitation of machine learning for the automated acquisition of knowledge and, on the other hand, the adoption of object-oriented representation models for easing the maintenance. In this context, adopting techniques for structuring knowledge representation in machine learning seems particularly appealing.

Inductive Logic Programming (ILP) is a promising approach for the automated discovery of rules in knowledge based systems. We propose an object-oriented extension of ILP employing multi-theory logic programs as the representation language. We define a new learning problem and propose the corresponding learning algorithm. Our approach enables ILP to benefit of object-oriented domain modelling in the learning process, such as allowing structured domains to be directly mapped onto program constructs, or easing the management of large knowledge bases.