Semantic relations of inclusion in some computational lexical models

Abstract

The relationships of inclusion (RI), hyponymy and meronymy, are analysed in two computer models; the SIMPLE project, a computer application for the treatment of the multilingual lexicon and the WordNet database. We have chosen these two applications because they have different approaches in the design of these relationships, but also can be noted to have something in common. On the one hand, WordNet represents a lexicographic model of lexical organization, where nouns are grouped into nets around a group of twenty-five semantic primitives that function as closed categories. On the other hand, the lexicographic proposal of the SIMPLE project is based on hierarchical and non-hierarchical conceptual relations where the RI are conditioned by the formal nature of the lexicon: semantic types and extended qualia structure (Pustejovsky, 1995, 2005).
Inclusion relations are explained taking into account the phenomena of categorization and composition of the meaning, which are associated with the logical definition of extension and intension. These relations organize the lexicon of a language through systems of inheritance hierarchies of meaning. Relations of sense are the basis for the conceptual model of computational lexical systems, so that for some scholars like Pustejovsky (1995) there is no distinction between the semantic and the computational conceptual approaches: both are synonymous. On the other hand, WordNet (Miller et al., 1993), a lexical database for English that has been the mainstay of online WordReference.com, the current multilingual dictionary, is based on the lexical memory models, which in turn are based on the semantic networks of Psycholinguistics.
Finally, in this work, we propose a model of lexical organization for the RI and we apply it to a group of words used in Spain, that belong to the field of computer security, which have been shed from its specificity and form part of the common language.