Concept mining from natural language texts

Abstract

WordNet dictionaries are a commonly used resource in the NLP field. Many papers discuss the theme of WordNet dictionaries, but the processes devoted to their automated construction have still much space for improvement. Many of them are based on machine translation strategies converting the dictionaries from one language to another. Most of them have one attribute in common: they use the knowledge about the language used in the form of dictionaries or even simple grammar parsers. In WordNet dictionaries, terms are represented in the form of concept hierarchies (hypernyms, hyponyms, ...). Since the ALOC approach was used to achieve similar concept structures, we can assume that it could also be used in the area of automated WordNet dictionary construction. This paper discusses such application of ALOC, in the form of correct assignment of new concepts into the existing WordNet hierarchy. © 2012 IEEE.