Back to first pageBack to first page Centre for Artificial Intelligence of UNL
Browse our site

CENTRIA news

PhD Public Defense of Martin SlotaThe public defense of Martin Slota's PhD Thesis will take place on
Tuesday, September 25th, at 14:00 in Sala de Actos (Ed IV) of FCT/UNL.

The jury has the following composition:

* Thomas Eiter (TU Vienna) - main examiner
* Andreas Herzig (CNRS) - main examiner
* João Pavão Martins (IST)
* José Júlio Alferes (UNL)
* Eduardo Fermé (U. Madeira)
* João Leite (UNL) - Martin's adviser

Title:
Updates of Hybrid Knowledge Bases

Abstract:
Increasingly many real world applications need to intelligently access,
combine and reason with large amounts of dynamically changing and highly
interconnected information. The vision of the Semantic Web is to foster such
advanced new applications by imbuing Web content with machine-processable
metadata. Throughout the last decade, two distinct knowledge representation
paradigms have been standardised to capture such metadata: ontology
languages based on Classical Logic and reasoning rules based on Logic
Programming. Both offer important features for knowledge representation and
the interest in their integration has recently resulted in frameworks for
hybrid knowledge bases that consist of an ontology and a rule
component. However, despite the abundance of work on dynamics of ontologies
and rules, when taken separately, evolution of hybrid knowledge bases has not
been addressed. The goal of this thesis is to tackle this problem, focusing on
updates of hybrid knowledge bases. A principled, formal understanding
of updates is particularly interesting as it paves the way towards automated
support for dealing with the vast amount of constantly changing information on
the Web.

In the first part of the thesis we develop two update semantics for hybrid
knowledge bases that fit the needs of particular use cases of hybrid knowledge
and provide the expected results when used in specific application domains.
The first semantics uses a given ontology update operator to update the
ontology component of a hybrid knowledge base in the presence of static
rules. Inspired be a realistic application, the second semantics offers a way
to modularly combine an ontology update operator with a rule update
semantics. It can be used for performing updates of hybrid knowledge bases
consisting of ontology and rule layers that share information through a
rule-based interface. Both of these developments constitute solutions to the
problem of hybrid updates for restricted classes of hybrid knowledge bases.

Subsequently, we seek to provide a general solution that could handle updates
of arbitrary hybrid knowledge bases. After pinpointing serious difficulties
due to the semantic incompatibilities between existing approaches to ontology
and rule updates, in the second part of the thesis we look for ways to bring
them closer together by developing semantic characterisations of rule
updates. We show that the classical semantic approach to updates, even when
applied to a very expressive semantic characterisation of logic programs,
leads to the violation of essential properties of existing rule update
semantics. This leads us to the development of richer semantic
characterisations of logic programs. The main result of this line of work is
the introduction of a generic method for specifying update operators that can
capture both classical updates and the historically first rule update
semantics. This constitutes the first common ground for these different update
paradigms and enables us to closely examine the semantic properties of rule
updates.

Together, the developments introduced in this thesis make it possible, for the
first time, to devise automated tools for supporting the dynamics of hybrid
knowledge, and at the same time provide a unifying perspective of the two
approaches to updates, seemingly irreconcilable, deepening our understanding
of the foundations of ontology and rule updates, taken separately and in
combination.

Keywords: Semantic Web, Knowledge Representation, Belief Change,
Ontologies, Logic Programs, Hybrid Knowledge Bases, Updates
Tuesday, 18th of September 2012

Centre for Artificial Intelligence of UNL
Departamento de Informática, FCT/UNL
Quinta da Torre 2829-516 CAPARICA - Portugal
Tel. (+351) 21 294 8536 FAX (+351) 21 294 8541

Fundacao_FCT