Arndt von Haeseler, Dominic Mainz, Indra Mainz, Ingo Paulsen, Katrin Weller - Knowledge Representation with Scientific Ontologies: Examples from Life Sciences
Theoretical considerations on the construction of ontologies as knowledge representation systems are currently a research focus examined with discipline-specific interests e.g. by computer sciences, information science, linguistics, artificial intelligence and philosophy. A growing interest in actually developing and applying ontologies is observed in the life sciences, especially in biology, biomedicine and bioinformatics. Among the most prominent bio-scientific ontology projects is the Gene Ontology (GO), which is also included amongst others in the Open Biomedical Ontologies (OBO), a collection of the National Center for Biomedical Ontology. We want to discuss general aspects of ontology engineering and aspects of using ontologies in life sciences. All scientific disciplines and also non-scientific areas of the World Wide Web demand for new types of (semantic) information integration solutions, in the field of life sciences integration of different data types is particularly needed. Besides commonly used data-formats such as texts and images, the diversity of specific biology data (e.g. nucleotide sequences, amino acid sequences, 3D structures of molecules and a manifold of other experimental results) requires structuring, integration and interrelation. Ontologies are currently designed to perform semantic annotation and integration. The Gene Ontology for example structures genes and information about them. The GO can be used as a unified vocabulary for describing genesis, this addresses the problem, that in genetics similar genes in different organisms are often not described consistently yet. Additionally to the management of the vast amount of biological data there is also an urgency to collect scientific cognitions and make them multi-disciplinarily accessible, e.g. for practical applications such as expert systems, data mining or information extraction. Current domain ontologies mostly do not make full use of the complex expressiveness offered by sophisticated ontology languages like OWL. Still, objections were raised that even such languages might not be extensive enough when it comes to encoding the complex semantics of biological knowledge. Specific problems for knowledge representation on the life science domain are rapid progress and thus permanent changes in terminology as well as different terminologies in use according to different scientific schools. We will have a closer look at existing bio-ontologies and their requirements on knowledge representation methods. We further give a short overview about our own experiences in constructing an ontology for bioinformatics tools and methods (BIO2Me). We are focusing on two aspects of knowledge modelling, that we have encountered during the ontology engineering process and that might also be of relevance for other domains: The modelling of concepts vs. individuals and vertical vs. horizontal hierarchy construction. The former aspect concerns the question of how to determine the smallest entities of a knowledge representation, the individuals. The latter one considers that options for horizontal structuring (included in the support of free-defined relationships between concepts) are hardly exploited in current ontologies. As classical knowledge representations are mainly working with hierarchies, particularly novices in ontology design still tend to model concepts in vertical structures. Both aspects are interrelated and thus regarded in their interaction.
References
Baclawski, K., Niu, T.: Ontologies for Bioinformatics. Cambridge Mass: The MIT Press, 2006.
Bernauer,J.: Analysis of part-whole relation and subsumption in the medical domain. –In: Data & Knowledge Engineering, 20, 405-415, 1996.
Bittner,T. & Donnelly, M. (2005): Computational ontologies of parthood, componenthood, and containment. – In: Proceedings of the 19thInternational Joint Conference on Artificial Intelligence, 382-387.
Cincinnati Children’s Pediatric Informatics: Bio-Ontologies. A List of Links. anil.cchmc.org/Bio-Ontologies.html (retrieved March 30, 2007).
Davies, J., Studer, R., Warren, P.: Semantic Web Technologies. Trends and Research in Ontology-Based Systems. Chichester: Wiley, 2006.
Fernández López, M.: Overview of Methodologies for Building Ontologies. In: Proceedings of the IJCAI-99 Workshop on Ontologies and Problem-Solving Methods (KRR5), 1999.
Gene Ontology Consortium: Gene Ontology. Tool for the Unification of Biology. Nature Genetics, 25: 25-29, 2000.
Handschuh, S., Staab, S.: Annotation for the Semantic Web. Frontiers in Artificial Intelligence and Applications, Vol.96, Amsterdam 2004.
Open Biomedical Ontologies, obo.sourceforge.net (retrieved March 30, 2007).
Stein, L.D.: Integrating Biological Databases. Nature Rev. Genet., 4(5):337–345, 2003.
Stuckenschmidt, H., van Harmelen, F.:Information Sharing on the Semantic Web, Berlin, Heidelberg 2005.
W3C Recommendation: OWL Web Ontology Language Overview, 2004,
www.w3.org/TR/owl-features/ (retrieved March 30, 2007).