A Pilot Study of the Characterization of English-Chinese Web Bilingual Data

摘要

With the rapid development of WWW, bilingual data appears on blogs, forums, etc. To the best of our knowledge, little work has been done to intensively study the characterization of bilingual data in English-Chinese mixed language pages. However, its distribution features do matters a lot for many text mining research topics, e.g. estimating training data quantity and quality for statistical machine translation. In this paper, we state several key issues to understand the characterizations of bilingual corpora on the web, and then we build an experimental platform to study the features of web bilingual data. Finally, we conduct the experiments and present the preliminary results.