Hypernym-Hyponym Acquisition-A Juxtapose Study

Relationship acquisition has always contributed as an important enhancement to Natural Language Processing. Using “definition extraction” methods proved to be a very useful approach, not only for relationship extraction but also for enhancing the limited coverage of words in WordNet. It also played a major role in developing question answer systems and ontology learning for unrestricted data. Various epistemologies developed for the later but it was noticed that many among the developed approaches suffered from low precision and low recall, especially the ones which were based on lexico syntactic patterns. This paper presents a comparative study of two different approaches used for Hypernym-Hyponym discovery. We start with analyzing the Pattern matching approach based on the Hearst Pattern and then compare it with LEILA which uses link grammar parser and deep syntactic analysis for relationship acquisition. Keywords— Hypernym, LEILA, Pattern Matching, Hearst Pattern, lexico-syntactic pattern