Self-enrichment of normalized LMF dictionaries through syntactic-behaviors-to-meanings links

Abstract

This paper reports on the status of an ongoing work to enrich the syntactic extension of normalized LMF dictionaries. It proposes an approach to find out the syntactic behaviors associated with the lexical entries and to link them to the corresponding meanings of these entries. The used corpus is constructed from texts associated with each meaning in the dictionaries, such as definitions and contexts. These texts are largely available and semantically controlled because of their association to the meanings, which fosters the establishment of efficient links. The proposed approach has been implemented and tested on an available Arabic normalized dictionary. The experiment concerned about 9,800 verbal entries. Both the identification of syntactic behaviors and the establishment of syntactic-behaviors-to-meanings links have been evaluated.