This article is part of the supplement: Proceedings of Ontologies in Biomedicine and Life Sciences (OBML 2010)
Proceedings
Grains, components and mixtures in biomedical ontologies
-
* Corresponding author: Ludger Jansen ludger.jansen@uni-rostock.de
- † Equal contributors
1 Institute of Philosophy, University of Rostock, 18051 Rostock, Germany
2 Institute for Medical Informatics, Statistics and Documentation, Medical University of Graz, Auenbruggerplatz 2, 8036 Graz, Austria
Journal of Biomedical Semantics 2011, 2(Suppl 4):S2 doi:10.1186/2041-1480-2-S4-S2
Published: 9 August 2011Abstract
Background
In biomedical ontologies, mereological relations have always been subject to special interest due to their high relevance in structural descriptions of anatomical entities, cells, and biomolecules. This paper investigates two important subrelations of has_proper_part, viz. the relation has_grain, which relates a collective entity to its multiply occurring uniform parts (e.g., water molecules in a portion of water), and the relation has_component, which relates a compound to its constituents (e.g., molecules to the atoms they consist of).
Method
We distinguish between four kinds of complex entities and characterize them in first order logic. We then discuss whether similar characterizations could be given in description logics, and finally apply the results to mixtures.
Results
At first sight, collectives and compounds seem to be disjoint categories. Their disjointness, however, relies on agreement about what are uniform entities, and thus on the granularity of description. For instance, the distinction between isomeric subtypes of a molecule can be important in one use case but might be neglected in another one. We demonstrate that, as implemented in the BioTop domain upper level ontology, equivalence or subsumption between different descriptions of same or similar entities cannot be achieved. Using OWL-DL, we propose a new design pattern that avoids primitive subrelations at the expense of more complex descriptions and thus supports the needed inferences.