DMCA
An Automatic Method to Identify Antonymy Relations
| Citations: | 2 - 0 self |
BibTeX
@MISC{Lucero_anautomatic,
author = {Cupertino Lucero and David Pinto and Héctor Jiménez-salazar},
title = {An Automatic Method to Identify Antonymy Relations},
year = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
Abstract. Although WordNet has been used in several applications in many fields of natural language proccessing, the fact that it is not a specialized lexical database (LDB) has lead to the building of LDBs for specific domains. Thus, it is very important to develop automatic methods for recognition of lexical relations embedded in natural language. In this paper we present a new method that identifies pairs of words in opposite relationships from a raw corpus format. The method was applied on pairs of related words obtained from a thesaurus created by Grefenstette’s method. We also used some features extracted from word contexts. These features were evaluated by means of the distance between words that appear in the same context, by lexical-syntactic patterns used to match regular expressions with word contexts, and by a lexical co-occurrence network built for each related word. The method was tested on a set of pairs of words taken from an economy corpus, and obtained a 80-percent accuracy, which is highly promising.
Keyphrases
automatic method antonymy relation related word word context 80-percent accuracy lexical co-occurrence network several application new method natural language proccessing opposite relationship regular expression raw corpus format specific domain lexical-syntactic pattern many field lexical relation grefenstette method economy corpus specialized lexical database natural language
