Designing effective web mining-based techniques for OOV translation

Abstract

Due to a limited coverage of the existing bilingual dictionary, it is often difficult to translate the Out-Of-Vocabulary terms (OOV) in many natural language processing tasks. In this paper, we propose a general cascade mining technique of three steps, it leverages OOV category to optimize the effectiveness of each step. OOV category based expansion policy is suggested to get more relevant mixed-language documents. OOV category based hybrid extraction approach is suggested to perform a robust extraction. A more flexible model combination based on OOV category is also suggested. Moreover, we conducted experiments to evaluate the effectiveness of each step and the overall performance of the mining technique. The experimental results show significantly performance improvement than the existing methods.