Systematising pre-editing methods and developing a staged editing support system for better use of machine translation

2021 Fiscal Year Final Research Report

• Project/Area Number: 19K20628
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 90020:Library and information science, humanistic and social informatics-related
• Research Institution: Nagoya University
• Principal Investigator: Miyata Rei 名古屋大学, 工学研究科, 助教 (70804300)
• Project Period (FY): 2019-04-01 – 2022-03-31
• Keywords: 前編集（プリエディット） / 機械翻訳 / 明示化方略 / 執筆支援システム / 制限言語 / 翻訳しにくい表現 / 翻訳品質

Outline of Final Research Achievements

We created Japanese pre-editing rules for improving the quality of machine translation outputs by collecting and analysing manual pre-editing instances. Our analysis suggested that explicitation strategies, which are used to explicitly indicate the content of source text, are effective to improve machine translation outputs. We also developed a prototype editing support system and implemented some of the created rules. This tool provides the following functions according to editing processes: detection of rule violations, suggestion of candidates for rewriting, ranking of the candidates, and automatic rewriting of the source text based on the selected candidate.

Free Research Field

図書館情報学

Academic Significance and Societal Importance of the Research Achievements

近年、機械翻訳の性能が向上し、各種の情報発信場面での利用が増えている。原文を事前に翻訳しやすい表現に書き換えることの有効性は指摘されながらも、その具体的な手法は十分明らかになっていなかった。本研究は、何をどう書き換えればよいかに関する具体的な指針を与えるだけでなく、ツールの提供を通じて、そのような書き換えに不慣れな書き手を支援するものであり、様々な利用者による機械翻訳の活用に貢献する。