Identification and Transformation of Comparative Sentences in Patent Chinese-English Machine Translation

Abstract

Machine translation for patent texts is a worldwide concern and more than 25% of the sentences in patent texts are comparative sentences. So the research on identification and transformation of comparative sentences in patent texts is very significant and can contribute to the patent Chinese-English machine translation. On the issue of identification of comparative sentences, there are some studies basically using statistical methods, but almost no study applies the rule-based method and no study uses the results of identification for Chinese-English machine translation. In this paper, based on HNC (Hierarchical Network of Concepts) theory, we applied the method based on rules to study the identification and transformation of comparative sentences in patent texts. Experiment shows us that the method can improve the performance of patent machine translation effectually.