Department of Chinese, Translation and Linguistics
Research Degree Forum
Automatic Evaluation of Machine Translation: The State of the Art
Presented by
Mr. WONG Tak Ming
PhD candidate, Department of Chinese, Translation and Linguistics, City University of Hong Kong
Date: 24 June 2010, Thursday
Time: 4:30 - 5:30 pm
Venue: B7603 (7/F, Blue Zone), Academic Building, CityU
Abstract
Over the last decade, the automatic evaluation of machine translation has been undergoing a paradigmatic evolution. Many novel methods have emerged to automate the evaluation process with considerably enhanced efficiency and also direct influence upon the development of machine translation itself. In particular, the BLEU metric has almost become a de facto standard in scoring the performance of a machine translation system, although, there are also various criticisms on both its theoretical and practical aspects. In this talk, I will provide an overview on the development of this kind of automatic evaluation measures in recent years, including their underlying assumptions, common approaches, and current application in different evaluation practices. Their merits and demerits will be assessed in a framework in connection with human evaluation and with the research and development of machine translation systems.
Speaker
Mr. WONG Tak Ming is currently a PhD candidate from the Department of Chinese, Translation and Linguistics. His research focuses on the evaluation of machine translation, in particular, its automated metrics and the theoretical studies of the relationship between human and automatic measures. On this basis, the study on proper selection and use of machine translation technology for lay users is also explored.
~ CTL Staff and Research Degree Students only ~