Lexicalizing Ontologies: The Issues Behind the Labels

Abstract

In information science, ontologies are used to capture knowledge about some domain of interest, by formally naming and defining the types, properties and interrelationships of the concepts that describe that domain. They are the building blocks of the Linked Data initiative in which datasets of related domains are linked to each other, and also to more general datasets, resulting in a huge space of interconnected data. It is precisely in this linking step that the linguistic descriptions used to name or label ontology entities (i.e., concepts, properties, attributes) become undeniably significant. This has also an impact in the subsequent process of ontology localization, thus turning it into a critical process. Based on our experience in the lexicalization and localization of several well adopted ontologies (FOAF, GoodRelations, the Organization ontology, among others) from English into Spanish, we propose a preliminary set of guidelines with a twofold goal. First, to guide users in the process of assigning labels and descriptions to ontology entities; second, to help terminologists and translators in the translation of these specific resources by providing them with coherent, user-friendly examples of how to apply the above mentioned guidelines