Soares Felipe, Rebechi Rozane, Stevenson Mark
Computer Science Department, University of Sheffield, Sheffield, UK.
Instituto de Letras, Universidade Federal do Rio Grande do Sul, Porto Alegre, Brazil.
Genomics Inform. 2020 Jun;18(2):e21. doi: 10.5808/GI.2020.18.2.e21. Epub 2020 Jun 15.
Scientific research is mostly published in English, regardless of the researcher's nationality. However, this growing practice impairs or hinders the comprehension of professionals who depend on the results of these studies to provide adequate care for their patients. We suggest that machine translation (MT) can be used as a way of providing useful translation for biomedical articles, even though the translation itself may not be fluent. To tackle possible mistranslation that can harm a patient, we resort to crowd-sourced validation of translations. We developed a prototype of MT validation and edition, where users can vote for that translation as valid, or suggest modifications (i.e., post-editing the MT). A glossary match system is also included, aiming at terminology consistency.
无论研究人员的国籍如何，科学研究大多以英文发表。然而，这种日益普遍的做法损害或阻碍了那些依靠这些研究结果为患者提供充分护理的专业人员的理解。我们建议，机器翻译（MT）可以用作一种为生物医学文章提供有用翻译的方式，尽管翻译本身可能并不流畅。为了解决可能危害患者的误译问题，我们采用众包方式对翻译进行验证。我们开发了一个机器翻译验证与编辑的原型，用户可以投票认为该翻译有效，或者提出修改建议（即对机器翻译进行译后编辑）。还包括一个词汇匹配系统，旨在确保术语的一致性。
