Improving Vietnamese WordNet using word embedding

Abstract

This paper presents a simple but effective method to improve the quality of WordNet synsets and extract glosses for synsets. We translate the Princeton WordNet and other intermediate WordNets to a target language using a machine translator, then the correct candidates are selected by applying different ranking methods: occurrence count, cosine similarity between words, cosine similarity between word embeddings and cosine similarity between Doc2Vec of sentences. Our approaches may be applicable to build WordNets in any language which has some bilingual dictionaries and at least a monolingual corpus in the target language.