Ontological Analysis for the Unification of Biology

The overwhelming amount of information stored in various data repositories including those available over the web emphasizes the relevance of knowledge integration methodologies and techniques to facilitate data sharing. The need for such integration has been already perceived for several years, but telecommunications and networking are quickly and dramatically changing the scenario. We implemented a wrapper translating from the XML ontology definition into LOOM, a formalism suitable for automatic classification [6]. The ontological analysis put in evidence the necessity of refining some assumptions made by the Gene Ontology developers. For example, metonymy is often used, since both enzymes and their functions are used in the same taxonomy. We are witnessing the unification of biology, since both biochemists and genetists now recognize a single universe of genes and proteins, and such unification is made possible also by the ever-increasing availability of the sequences of entire genomes. Our experience has proved that the ontologies produced by means of the ONIONS methodology support: • formal upgrading of terminology systems: term classification and definitions are now available in a common, expressive formal language; However, such a growing demand has to rely on a solid conceptual foundation in order to give a precise semantics to the terabytes available in different genome databases and eventually traveling over the networks. The actual demand is not for a unique conceptual izat ion, but for an unambiguous communication of complex and detailed concepts (possibly expressed in different languages), l eav ing each user f ree to make expl ic i t h i s /her conceptualization. • conceptual explicitness of terminology systems: (local) term definitions are now available, even though the source does not include them explicitly; • ontological comparability, since pre-existing ontology libraries pertaining to different fields are largely employed. In conclusion, we point out the following important features of ontologies: • Semantic explicitness. Often this task is not an easy one to be achieved, since a deep analysis of the structure and the concepts of terminologies is needed. Such analyses can be performed by adopting an ontological approach for representing terminology systems and for integrating them in a set of ontologies. • An explicit taxonomy. • Explicit linkage to concepts and relations from generic theories. • Absence of polysemy within a given formal context. • Modularity of contexts. • Some minimal axiomatization to detail the difference among sibling concepts. Ontologies not only make knowledge re-use easier, they are also the foundation of standardization efforts since they make explicit the conceptualizations behind a terminology or a model. Therefore, an ontological analysis of a domain, which results into defining a library of ontologies, consist in the explicitation of implicit relationships among concepts in a given context. • A good naming policy. • Rich documentation.