翻译公司E2F和机器翻译技术公司Lilt合作的首个自适应机器翻译大规模应用案例

 

 

 

 

美国加利福尼亚州翻译公司E2F本周发布成功案例，该公司15年来最大的单次翻译项目：十天内把177万字翻译成6种语言。此任务由有超过100个翻译和编辑的E2F团队以及自适应机器学习（ML）技术为基础的公司Palo Alto-based Lilt, Inc共同完成。

 

 

Lilt公司结合自适应机器学习（ML）的机器翻译技术，创造了机器援助翻译的新典范。该系统通过经验，智能和人的意见进行学习，通过共同合作，提出建议并随着时间而改进准确度及提高翻译效率。

 

结果是，译者花费小部分时间和成本的审阅获得几乎相同的翻译质量。这证明，机器的协助可以为客户节省采用传统人力翻译服务的一半（或更多）的时间及费用。

 

 

Lilt公司CEO， Spence Green，在爱尔兰都柏林举行的LocWorld31公布了该案例的结果，并说道： “这是有史以来第一次翻译团队与机器翻译系统集体训练，及实时性交互。该项目证明了自适应机器翻译技术具备大规模生产应用”。

 

 

 

e2f and Lilt Case Study: First Large-Scale Application of Auto-Adaptive Machine Translation

 

 

From e2f

 

 

 

Combining Machine Translation (MT) with auto-adaptive Machine Learning (ML) enables a new paradigm of machine assistance. Such systems learn from the experience, intelligence and insights of their human users, improving productivity by working in partnership, making suggestions and improving accuracy over time.

 

The net result is that human reviewers produce far higher volumes of content, with nearly the same level of quality, for a fraction of the time and cost. Machine assistance can save customers up to one half (or more) of the price of traditional high-quality human translation services. Or, if you’ve been used to machine translation alone and have been unhappy with the results, watch your translation quality rise dramatically with a marginal increase in price.

 

Case Study: Travel Portal Translation

 

A large travel and tour web site wanted to localize 1.77 million words of content from their catalog into 6 languages within a two week window. The hard deadline was to be ready to accommodate the summer vacation plans of millions of global users with more destinations and new activities. Successfully achieving this goal required rapid mobilization of a high-quality team of humans, fully-empowered by robust machine assistance technology.

 

e2f, based out of San Jose, California, with over 15 years of success in the translation and localization business, provided the “human capital” for the project. Their team was comprised of 100+ experienced translators, editors, and reviewers, plus seven project managers and a senior localization engineer.

 

Lilt, based out of Palo Alto, California, provided the translation engine for the project. Founded in 2015, its technology platform incorporates the latest research in Natural Language Processing (NLP), Human-Computer Interaction (HCI), and Machine Learning (ML).

 

The Lilt platform proved invaluable in augmenting e2f’s human staff, increasingly translation speeds far beyond the industry average of 335 words per hour. The automation process required transformation of source Excel documents into a format suitable for automated processing, which were then uploaded into the correct accounts in Lilt via scripts calling Lilt’s APIs. Once translations were made within the Lilt system, output was generated and transformed back into Excel documents in the target languages.

 

Lilt-e2f-API

 

e2f’s implementation of Lilt utilized Lilt’s API, plus pre- and post-translation

processing and quality checking

 

Human translators could then accept or revise these segments. If Lilt’s suggestions were rejected, the approved human translations were fed back into the system, which learned from the human’s perspective and expertise. This positive feedback loop enabled faster and more accurate translation over time as the human translators contextually taught the system preferred translations of terms and phrases.

 

The client’s reaction was overwhelmingly positive. The number of errors was low compared to traditional machine translation solutions and the quality in line with standard human translations. Given a two-week window, the project was actually completed within 10 days.