Francesca A. Lisi

Francesca A. Lisi

ESWC 2007

tutorial on

Inductive Logic Programming Approaches to

Ontology and Rule Acquisition

for the Semantic Web




Abstract


Ontology and rule languages for the Semantic Web are logically founded on fragments of first-order logic, namely Description Logics (DLs) and hybrid DL-Horn Clausal Logics. Building ontologies as well as rules on top of ontologies is a very demanding task also from the Knowledge Acquisition viewpoint. When performing this task, Semantic Web practitioners could take benefit from the application of Machine Learning (ML) methods and techniques. The ML approach known as Inductive Logic Programming (ILP) seems to be particularly appropriate due to the common roots with computational logic. The tutorial on Inductive Logic Programming Approaches to Ontology and Rule Acquisition for the Semantic Web will provide a survey of ILP proposals for learning with DLs and hybrid DL-Horn Clausal Logics. Attendees are expected to be knowledgeable in (quasi-)standard languages for ontologies and rules for the Semantic Web (OWL and SWRL) and to have basic notions of first-order logic. The tutorial will focus mainly on theoretical aspects but the presentation will be rich in illustrative examples, animations, etc. in order to assure that the attendees become acquainted with the logical foundations of ontology and rule languages and fully comprehend the potential of ILP for the Semantic Web.
[back to top]


Description


Aims

The tutorial is intended to provide a survey of research in Inductive Logic Programming restricted to the literature of interest to the Semantic Web area, namely ILP proposals for learning with DLs and hybrid DL-Horn Clausal Logics which can contribute to solve a hard problem in the Semantic Web: (semi-)automated acquisition of ontologies and rules. It covers topics like
in equal depth.

Attendees are expected to be knowledgeable in (quasi-)standard languages for ontologies and rules for the Semantic Web (OWL and SWRL) and to have basic notions of first-order logic.

Content

The layered architecture of the Semantic Web poses several challenges in the field of Knowledge Representation and Reasoning (KR&R). E.g., the design of OWL for the ontological layer has been based on DLs, more precisely on the DL SHIQ (HorrocksP-SvH03). Also SWRL, recently proposed for the logical layer, extends OWL 'to build rules on top of ontologies'. It is intended to bridge the notorious expressive gap between DLs and Horn clausal logic (or its fragments) (Borgida96) in a way that is similar in the spirit to hybridization in KR&R systems (Donini98, Rosati05).
Building ontologies as well as rules on top of ontologies is a very demanding task also from the viewpoint of Knowledge Acquisition. When performing this task, Semantic Web practitioners could take benefit from the application of ML methods and techniques. The ML approach known under the name of ILP seems to be particularly promising due to the common roots with computational logic. ILP has been historically concerned with concept learning from examples and background knowledge within the representation framework of Horn clausal logic and with the aim of prediction. More recently ILP has moved towards either different first-order logic fragments (e.g., DLs) or new learning goals (e.g., description). Inducing DL concept descriptions from examples has been attacked mostly by heuristic means (CohenBH92,CohenH94,KietzM94) and very recently in a formal manner (BadeaNC00). Learning in hybrid DL-Horn clausal logics has very recently attracted attention in the ILP community. In (RouveirolV2000) the
chosen language is Carin-ALN, therefore example coverage and subsumption between two hypotheses are based on the existential entailment algorithm of Carin. Following (RouveirolV2000), Kietz studies the learnability of Carin-ALN, thus providing a pre-processing method which enables ILP systems to learn Carin-ALN rules (Kietz03). In (Lisi05), Lisi proposes a general framework for learning in AL-log.

Presentation style

The tutorial will focus mainly on theoretical aspects but the presentation will be rich in illustrative examples, animations, etc. in order to assure that the attendees become acquainted with the logical foundations of ontology and rule languages and fully comprehend the potential of ILP for the Semantic Web.
[back to top]


Schedule


First part: Logical Foundations of Ontology and Rule Languages for the Semantic Web (1h)

    1.  Description Logics (DLs) (BaaderCMcGNPS03)

Second part: Inductive Logic Programming with DL-based representations (2h)

[back to top]


Bibliography

[back to top]