Squeezing lemon with GATE
Résumé
An increasing number of enterprises are beginning to include ontologies into Text Analytics (TA) applications. This can be challenging for a TA group wishing to avail of such technologies due to the manual effort needed to map language resources within a TA system for a new domain. Ontology lexicalization offers a solution to this problem by seeking to automatically generate lexical resources in order to shrink the manual effort of this conceptto-text mapping process. However, conventional approaches are limited in that they often can only generate term mentions of proper noun, personal noun or fixed key phrases from concept labels in ontologies. Such approaches do not generalize to cope with more complex concept mentions such as nominal compounds or multi-word expressions. An alternative consideration is lemon-Lexicon Model for Ontologies which offers a more sophisticated solution to this problem. We describe a simple use case for exploiting lemon within a widely used open-source TA framework and demonstrate how lemon generated lexical resources are at least comparable in agreement to OntoRootGazeteer, a conventional ontology lexicalization approach.
Origine : Fichiers produits par l'(les) auteur(s)