25 2 1

25 2 1

i

ii

Abstract Supporting Collaborative Translation with Text Summarization iii Yuri HORITA In recent years, there is an increasing demand for translation with the background of globalization. General methods of translation are machine translation and expert translation. However, because of the balance of translation quality and costs, we can not meet growing demands for translation only with these methods. Therefore, collaborative translation, which is the method to improve the translation quality by several translators, has been proposed. The proposed method costs lower than translation by experts and can achieve a high quality translation than machine translation. In collaborative translation, a translator makes translation of original sentence or improve the translation made by other translator. In addition, as a tool to support it, the collaborative translation support tool exists. This is intended to support translators to do collaborative translation over the Internet. However, because it is not reviewed by the users and not improved based on it, the tool is still incomplete. For improvement of the tool, the issues are as follows. Finding the problems of collaborative translation support tool based on the analysis of user behavior using the tool The collaborative translation support tool is not evaluated after the development and the problem of this tool is not clear. Analyzing the behavior of the users by using a method such as user reviews and finding the problems of this tool are needed. Since there could be a simple matter of implementation, what is causing must be found. The design and implementation of the new system to solve the problem that necessary information for translation can not be obtained Among the problems that have been classified, I must resolve those of high importance on a priority basis. There is a need to design and implement a new system for it.

iv In this study, I discover the problems of collaborative translation support tool, implement a new system to resolve the problems, and evaluate it in the end. Firstly, I evaluate the tool by user review and list up problems. I get feedback from users using this tool to a real translation demand. Secondly, I classify the problems into the functional problem, interface problem, and so on. I reveal the priority problems to be solved and causes of them. Then I design a new system to solve these problems. Finally, I design and implement a new system. Many problems of the tool have revealed in the experiment. There was the high priority problem which is not just an implementation problem. It is the problem of the amount of information. For this the problem, I tried to resolve by presenting a summarization of the original sentences. Users can get necessary information for translation from summarization easier than reading all of the original sentences.i designed the new tool with text summarization and implemented it. Contributions of this study are as follows. Finding the problems of collaborative translation support tool based on the analysis of user behavior using the tool By evaluation of the tool by user review I found some problems of the tool. In addition, I found that there are two fundamental problems. One is that the amount of sentences is excessive for the users to read in order to obtain the amount of information required for translation. The other is that it difficult for them to select a task from many tasks that are presented to them. The design and implementation of the new system to solve the problem that necessary information for translation can not be obtained In order to solve the problem that it is difficult for the users to obtain the required information for translation because of the amount of sentences to read to obtain the context information is excessive, I have designed and implemented a new interface with text summarization.

1 1 2 3 2.1....................................... 3 2.2............................... 5 2.3.................................... 7 3 8 3.1.......................................... 8 3.2............................. 9 3.2.1 Trados Google Translator Toolkit.............. 9 3.2.2................... 11 4 11 4.1................................. 12 4.2................................. 14 4.3..................... 15 5 16 6 19 6.1..................................... 19 6.2.................... 21 7 21 23 23

1 SNS 1

[1] [2] [3] 2 3 2

4 5 6 7 2 2.1 [2] [3] 3

1 1: 4

Lin [4] Crowd-Assisted Translation(CAT) [5] 2 CAT 2.2 Web Amazon Mechanical Turk 1) Web 1) Amazon Mechanical Turk(http://www.mturk.com/) 5

件に大きく左右される 具体的な例として Amazon Mechanical Turk では作 業者にアメリカ在住の大学生が多く 彼らが長期休みに入っている八月には若 い作業者の数が多くなり 彼らの新学期が始まる十月にはその数が減るという 調査結果が存在している [6] 母集団の人数や年齢分布のぶれ等をどのように制 御するかに関する研究が行われている また クラウドソーシングは翻訳という分野においても利用され始めている 例えば 日本国内においては Conyac1) というソーシャル翻訳サービスが存在す る 依頼者が原稿を Conyac に委託すると クラウドソーシングによって集め られた作業者に対しその原稿の翻訳タスクが提示される 翻訳を行うことがで 図 2: CAT による翻訳プロセス 1) Conyac(http://www.conyac.cc/) 6

Conyac [7] [1] 2.3 (Language Grid) 1) [8] (accessibility) 1) (http://langrid.org/) 7

3 Toolbox 3.1 Web Language Grid Toolbox 1) [9] Toolbox Toolbox [5] Web 3 3: Kyoto University is 1) Language Grid Toolbox(http://langrid.org/tools/toolbox/) 8

located on Tokyo University road. 1 Tokyo University road Higashioji St. Kyoto University is located on Higashioji St. 2 1 1 on along Kyoto University is located along Higashioji St. 4 4 3.2 Trados Google Translator Toolkit 3.2.1 Trados Google Translator Toolkit Trados 1) Microsoft Word Trados 1) Trados(http://www.trados.com/) 9

1 10 Google Translator Toolkit 1) Google 2) Web Web Wikipedia 3) Google Google 4: 1) http://translate.google.com/toolkit/ 2) http://translate.google.com/ 3) http://en.wikipedia.org/ 10

2013 1 Google Google 3.2.2 1: Trados Google Translator Toolkit 1 1 1 4 11

4.1 12

Save 1 There is no translation yet. Save 1 2 1 13

4.2 1 UTF-8 14

4.3 15

5 [10] 5 6 GUI GUI 16

5: 17

6: 18

6 6.1 19

20

6.2 Language Grid Toolbox Toolbox Toolbox 4 Toolbox 7 21

1. 2. 22

[1] Zaidan, O. and Callison-Burch, C.: Crowdsourcing translation: Professional quality from non-professionals, Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, pp. 1220 1229 (2011). [2] Little, G., Chilton, L., Goldman, M. and Miller, R.: Exploring iterative and parallel human computation processes, Proceedings of the ACM SIGKDD workshop on human computation, ACM, pp. 68 76 (2010). [3] Douglas, S. and Craig, C.: Collaborative and iterative translation: an alternative approach to back translation, Journal of International Marketing, Vol. 15, No. 1, pp. 30 43 (2007). [4] Lin, D., Murakami, Y., Ishida, T., Murakami, Y. and Tanaka, M.: Composing human and machine translation services: Language grid for improving localization processes, Proceedings of the Seventh International Conference on Language Resources and Evaluation (2010). [5],,,,,,, : (2012). [6] Ross, J., Irani, L., Silberman, M., Zaldivar, A. and Tomlinson, B.: Who are the crowdworkers?: shifting demographics in mechanical turk, Proceedings of the 28th of the international conference extended abstracts on Human factors in computing systems, ACM, pp. 2863 2872 (2010). [7] Morita, D. and Ishida, T.: Collaborative translation by monolinguals with 23

machine translators, Proceedings of the 14th international conference on Intelligent user interfaces, ACM, pp. 361 366 (2009). [8] Ishida, T.: Language grid: An infrastructure for intercultural collaboration, Symposium on Applications and the Internet 2006 (SAINT-06), IEEE/IPSJ, pp. 5 100 (2006). [9] Tanaka, M., Murakami, Y., Lin, D. and Ishida, T.: Language Grid Toolbox: Open source multi-language community site, Universal Communication Symposium (IUCS), 2010 4th International, IEEE, pp. 105 111 (2010). [10],, : (2003). 24