Extracting new relations to improve ontology reuse
Keywords:aligment, information extraction, ontology
AbstractOntologies reuse in biomedicine faces several challenges, which include the complexity of the domain and the articulation of different vocabularies, many of them with thousands of terms. Those ontologies are still subject to changes to improve their quality and to solve existing deficiencies. Among them, we highlight the problem of concepts that should be related, according to their definition, but are not explicitly connected through relations in the ontology. This could hinder ontology comprehension and limit the scope of the domain represented by the vocabulary, which can eventually encourage the development of new ontologies instead of reusing existing ones. Due to these issues, the adoption of ontology tools to support the discovery of implicit relations, intra and inter ontologies, is strongly recommended. Although there are tools geared to find new relationships in ontologies, such tools fail to consider the specific rationale used to organize the knowledge related to a specific domain. This paper proposes an approach for information extraction in ontologies, which uses the definition and nomenclature of each concept to extract implicit information that can complement the knowledge they contain. This increase in knowledge allows an increase in the quality and semantics of ontologies, and thus improves several processes, including the reuse. We have applied the approach to the biomedical domain and, as a result, we have discovered a set of possible new relations for a single ontology as well as relations between two different ontologies.
Download data is not yet available.