Francesca A. Lisi
Dipartimento di Informatica - Università degli Studi di Bari
abstract in pdf slide in pdf
Learning Rules on top of Ontologies: An Inductive Logic Programming Approach
New challenging application areas like the Semantic
Web require the definition of rules on top of ontologies.
Hybrid systems for Knowledge Representation and Reasoning (KR&R) that combine
(fragments of) Horn clausal logic and Description Logics, notably AL-log, have been
invented to provide one such unified framework for dealing with both relational and
structural data. Yet acquiring hybrid rules is a demanding and time-consuming task.
It can be (partially) automated by applying Machine Learning algorithms, more
precisely those ones following the approach known under the name of Inductive Logic Programming (ILP).
In this seminar I shall present a general framework
for learning rules on top of ontologies that adopts the methodological apparatus of
ILP within the KR&R setting of AL-log. Also I shall briefly discuss an ILP system
that implements an instantiation of the framework and the preliminary results of its
application in the Semantic Web context.