Student Award | News

4th AAMT Nagao Award Student Award

Dr.John Richardson, Graduate School of Informatics, Kyoto University

Awarded Paper:
"Improving Statistical Machine Translation with Target-Side Dependency Syntax”
(This was submitted as a doctoral dissertation at Kyoto University in 2016 under the supervision of Prof. Sadao Kurohashi.)

Reasons for conferring the award:
This thesis proposes a novel Tree-to-Tree (T2T) translation framework and dependency tree language model exploiting target language dependency syntax. These are applied to reranking translation candidates, reordering with flexible non-terminal symbols, and post-editing of function words. These methods demonstrate the potential to improve translation quality for multiple language pairs, particularly by employing Recurrent Neural Network language models to rerank translation candidates and determine insertion positions for non-terminal symbols. The software behind this research has been released as the translation system KyotoEBMT.
This work represents the peak of many years of research into Tree-to-Tree Machine Translation and Example-Based Machine Translation (EBMT). Furthermore, it suggests possible future directions for new research into T2T/EBMT making use of the latest technologies of Neural Machine Translation.


3rd AAMT Nagao Award Student Award

Awardee: Akiva Miura Nara Institute of Science and Technology - the second semester of the doctroal program
Awarded Paper: Accuracy enhancement in multilingual machine translation that uses pivot language model (Under the advise of Dr. Graham Neubig, this was summarized as a dissertation at Nara Institute of Science and Technology in year 2016.)
Reason of Awarding: In this disseration, Miura suggests a new realization method to use the third language as a pivot language when the quantity of corpus for translating languages is not enough. And he succeeds to indicate that it can achieve the great enahancement in translation accuracy. We acknowledge this excellent dissertation, indicating the possibility of future development and new research direction about translation with pivot language, as it deserves the AAMT Nagao Award Student Award. Also, it is expected for the future that the application to the language pairs with big differences on wording order such as for translating Japanese and other Asian languages via English.


2nd AAMT Nagao Award Student Award

Awardee: Isao Goto Graduate School of Informatics, Kyoto University (Currently at NHK Science & Technology Research Laboratories)
Awarded Paper: Word Reordering for Statistical Machine Translation via Modeling Structural Differences between Languages (This was polished as fellowship dissertation for 2014 Graduate School of Informatics, Kyoto University under the direction of Dr. Sadao Kurohashi.)
Reason of Awarding: In this dissertation, Goto suggests three methods to improve the accuracy in word reordering for Statistical Machine Translation with language combination whose word order is greatly different such as for Japanese and English. Using the corpus for patent translation, he confirms the effectivity over huge test translations from Japanese to English, and from Chinese to English. His first method actualizes the modeling, to discriminate word reordering, which considers the sentence structure of source language for the translation based on phrase. We acknowledge the excellence in improving the translation accuracy without using the syntax analyzer. With his second method, we acknowledge his unconventional idea to apply the ITG syntax analysis in the post-word reordering translation of the non-head postposition language converted into another language. With his third method, the syntax analyzer for the source language is structured by the use of the target language's syntax analyzer which projects the sentence structure of target language against the sentence structure of source language. This method actualizes the high accuracy in pre-word reordering translation. We acknowledge the superiority as it can be widely applied for the translation into the language, such as English, with which the highly accurate syntax analyzer can be used, no matter whatever the source language may be. As just described, Goto's brilliant research which considers empirically and deeply to the word reordering for Statistical Machine Translation. Thus we acknowledge that he deserves our AAMT Nagao Student Encouragement Award.


1st AAMT Nagao Award Student Award

AAMT Nagao Award Student Award has been newly established since this year, and the first AAMT Nagao Award Student Award was awarded to the following person.

Awardee: Rei Miyata University of Tokyo, Graduate School of Education, Ph.D First Grade
Awarded Paper: The awarded paper is the one written as a Master’s thesis of 2013 University of Tokyo, Graduate School of Education, under the guidance of Professor Kyo Kageura.
“A Study on Framework and System Environment to Support Multilingualization of Municipal Website Documents ”
Reason of Awarding: This paper proposes a controlled language and an authoring environment required for multilingual translation that takes advantage of machine translation technologies, for procedural documents of municipalities as a target. Although there have been lots of researches on controlled languages and their effect verifications to machine translations, this paper is highly evaluated in the point that it proposes an original methodology referring as far as document structure and creating process, from the library and information science point of view. Since this paper is a theoretical and empirical excellent research report relating to a new system and services related to machine translations, it is considered to be worthy of the AAMT Nagao Award Student Award.
Selection committee chairman: Masaaki Nagata (NTT)
Selection committee: Hirokazu Suzuki (Toshiba Corporation) Masaru Fuji (Fujitsu)
Recommender: Kyo Kageura (Graduate School of Education, University of Tokyo ) Tony Hartley (Tokyo University of Foreign Studies)


Last Updated: 19 Aug. 2017