Toward automatic construction of lexical semantic networks

Nowadays, language resources, such as machine readable dictionaries, WordNets, thesaurus, ontology, and semantic networks, are becoming crucial bases for many intelligent systems like Machine Translation, Information Retrieval, and Natural Language Understanding, etc. The purpose of this work is to extract features and to analyze the manually built Korean wordnet, namely U-WIN, for the automatic construction of a feature based U-WIN. We used ICA and CBC clustering methods and methodologies in the experiments and compared their results. Further we have identified various issues affecting the quality of clustering score.