This publication advances the state-of-the-art in ontology learning by presenting a set of novel approaches to the semi-automatic acquisition, refinement and evaluation of logically complex axiomatizations. It has been motivated by the fact that the realization of the semantic web envisioned by Tim Berners-Lee is still hampered by the lack of ontological resources, while at the same time more and more applications of semantic technologies emerge from fast-growing areas such as e-business or life sciences. Such knowledge-intensive applications, requiring large scale reasoning over complex domains of interest, even more than the semantic web depend on the availability of expressive, high-quality axiomatizations. This knowledge acquisition bottleneck could be overcome by approaches to the automatic or semi-automatic construction of ontologies. Hence a huge number of ontology learning tools and frameworks have been developed in recent years, all of them aiming for the automatic or semi-automatic generation of ontologies from various kinds of data. However, both the quality and the expressivity of ontologies that can be acquired by the current state-of-the-art in ontology learning so far have failed to meet the expectations of people who argue in favor of powerful, knowledge-intensive applications based on logical inference. This work therefore takes a first, yet important, step towards the semi-automatic generation and maintenance of expressive ontologies.
Expressive ontologies are an important prerequisite for a growing number of knowledge-intensive applications nowadays emerging from semantic web research. These applications, many of them originating in fast growing areas such as bioinformatics or medicine, demand for rich axiomatizations of huge and conceptually complex domains of interest. However, the sheer size and complexity of the required ontologies make high demands on the modeling capabilities of human ontology engineers, and modeling errors as well as unforseen logical consequences frequently hinder the practical application of the acquired knowledge. But even though any kind of automatic support in the acquisition and evaluation of expressive ontologies would be highly desirable, ontology learning so far has largely concentrated on the generation of relatively lightweight ontologies.
This PhD thesis presents a set of novel approaches to facilitate the semi-automatic generation and maintenance of expressive OWL DL ontologies. It describes the benefits and applications of logically complex axiomatizations while at the same time highlighting some of the greatest challenges for future ontology learning – including aspects of user interaction as well as uncertainty and inconsistency handling. The methods proposed by this thesis aim to combine the strengths of lexical and logical knowledge acquisition techniques. Their implementations, based on machine learning, natural language processing, formal concept analysis and OWL reasoning, demonstrate the practical feasibility of automatic support in engineering expressive ontologies. Particular emphasis is put on various aspects of ontology quality such as formal correctness, logical consistency and completeness.
Altogether, this thesis takes a first, yet important, step towards the semi-automatic generation of logically complex axiomatizations. Our experiments give raise to the hope that we can finally widen the knowledge acquisition bottleneck for reasoning-based applications.
IOS Press, Inc.
6751 Tepper Drive
Clifton, VA 20124
Tel.: +1 703 830 6300
Fax: +1 703 830 2300 firstname.lastname@example.org
(Corporate matters and books only) IOS Press c/o Accucoms US, Inc.
For North America Sales and Customer Service
West Point Commons
Lansdale PA 19446
Tel.: +1 866 855 8967
Fax: +1 215 660 5042 email@example.com