"lemon: An Ontology-Lexicon model for the Multilingual Semantic Web"

In order to allow ontologies to interact with multilingual text in both the analysis and the generation mode, it is necessary to model the relation that natural language expressions have with language-independent knowledge representation systems. Most of the latter use a label attribute to encode the natural language expressions that correspond to a concept. And often, such labels exist only in English, or only in the language of the country for which a taxonomy or ontology has been designed. Such labels correspond in fact to terms, which are not explicitly linked to other terms/labels of other concepts. As such, a lot of information about possible linguistic realizations of concepts is left out. The Semantic Web, and in particular the Linked Data project, proposes solutions that allow for the re-use of lexical and terminological resources by semantic interlinking. However, currently, there is no standard for describing the relationship between natural language expressions and ontology elements. Therefore a central aspect in the Monnet project on ‘Multilingual Ontologies for Networked Knowledge’ (http://www.monnet-project.eu/) is in the design and development of a model that associates linguistic information with domain semantics. This model, which we call ‘lemon’ (lexicon model for ontologies), is built on existing work, in particular LMF, ISOcat, SKOS, LexInfo and LIR. Lemon is an RDF model that allows for lexical data to be shared and interlinked on the Web and is a central endeavor towards standardizing lexicalized, multilingual knowledge representation on the Semantic Web.