Constructing Vietnamese WordNet: A Case Study

WordNets are commonly used in tasks such as summarizing documents, extracting information, translating and creating other lexical resources. This paper presents experiments in constructing a Vietnamese WordNet (VWN) from a variety of freely published resources in several languages. The VWN has the same structure as the Princeton WordNet. Our algorithm translates several existing WordNets to Vietnamese using a freely available machine translator, removes translation ambiguities by applying ranking methods based on occurrence counts and Google distances on translation candidates. We also establish connections between synsets and extract glosses for synsets. Finally, we carefully look at the VWN created and identify problematic issues in the VWN due to differences in culture and agglutinative morphology of Vietnamese and other languages used.