![]() Special thanks should be given to Berta González Saavedra for building Latin Vallex and to Christophe Onambele for both providing helpful remarks and computing the results shown in Table 2. In particular, in the context of the exploitation of the linguistic resources for ancient languages built over the last decade, we compute and evaluate the overlapping between a selection of homogeneous lexical subsets extracted from two lexical resources for Latin. In order to apply and evaluate our method, we propose a normalized coefficient of overlapping that measures the overlapping rate between a valency lexicon and a WordNet. As for the latter, our work wants to contribute to the research task dealing with merging lexical resources. As for the former, we wonder if there are lexical classes that "impose" themselves regardless of the fact that they are explicitly recorded as such in source lexical resources. ![]() This is motivated by both theoretical and practical reasons. In this paper, we present a method for measuring the degree of similarity between a valency-based lexical resource and a WordNet. However, all of them deal with lexical items as common basic components, which are described according to criteria that may vary from one resource to another. Different lexical resources may pursue different views on lexical meaning.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |