How are global businesses using the latest Neural techniques to prioritise their content for Translation?
Collecting feedback from users isn’t innovative in itself, but combining it with a raw MT approach…
Neil Campbell, Lingo24’s Director for Sales and Marketing, explores how Neural Machine Translation is unlocking more opportunities, like localization at scale, for global enterprises.
Machine Translation (MT) has been the hottest topic in the language industry for many years as an exciting technology with the potential to democratise how we publish and consume content, on a global scale.
Yet, the first waves of statistical MT output failed to reach the peak of the early marketing hype.
But MT in the Neural age has come a long way: the state of the art in MT has been transformed through the computational power of modern Graphical Processing Units (GPUs).
Neural MT is also now underpinned by architectures that can deliver greater fluency and accuracy. Highly performant and impactful neural engines can also now be customised with much less domain specific, organisational data.
It’s now unlocking more use cases than ever, presenting a viable option for global enterprises that need to accelerate their workflows and localize at scale.
Neural MT is now unlocking more use cases than ever, presenting a viable option for global enterprises that need to accelerate their workflows and localize at scale.
Read about real life examples of global enterprises that are taking advantage of Neural MT, including global publisher Elsevier: download our Point of View paper.
For global organisations looking to localize their content at scale, either at high volumes or throughputs (e.g. millions of products a week, thousands of user-generated comments an hour etc.), neural MT is now a core requirement of their localization architecture. The other is automation.
Traditional translation approaches simply can’t meet the demands of such scale. This is why many larger enterprise customers now rely on neural MT-accelerated workflows to support their growing demands, either on an ongoing basis or to deliver time-sensitive projects. Many larger enterprise customers now rely on Neural MT-accelerated workflows to to localize at scale.
Many larger enterprise customers now rely on neural MT-accelerated workflows to to localize at scale.
Using a high-quality engine enables translators to make a step change in their productivity. It helps focus the expertise and nuance that a human linguist brings on challenging or high-risk content, or getting a brand’s desired tone of voice or values just right.
But, the true power comes in where a neural MT engine is capable of real-time adaptation: where it can learn from the real-time edits of translators performing the post-editing activity. These improvements are no longer queued up waiting for the next point when the engine is retrained, which can often be many months later, if at all.
Using Neural MT as an accelerant was the only viable solution for Elsevier, a leading global publisher of scientific, technical and medical content.
We helped them translate over one million words in twelve weeks to support the launch of their ClinicalKey Student app for medical students in two key markets. It had to be live in time for the new academic year.
Our customised approach to neural MT met Elsevier’s needs for a quick turnaround, without compromising quality. With human-only translation this volume of content would have required 10 months and the app would not have been delivered in time.
With human-only translation this volume of content would have required 10 months and the app would not have been delivered in time.
But it’s not just localization at scale. The power of Neural MT is unlocking more use cases than ever for global enterprises, including data-driven translation, user generated content and multilingual artificial intelligence. Read about the 7 use case neural MT delivers: download our Point of View paper.