Update October 9th 2019: This blog refers to adaptive/statistical machine translation, which has now been replaced by SDL’s powerful neural machine translation (NMT) solution. To learn more about NMT please click here.
In recent years we have seen an explosion in digital data driven by technology’s capacity to store an increasing amount of information and by a rise in the number of people connected to the internet. In today’s world we’re seeing even more interconnectivity between people and devices, which will only perpetuate this upward trend in digital data. The IDC predicts that there will be a staggering 300 times more data in 2020 compared to 2005¹. Currently, the localization industry has limited resource to keep pace with all of the content that needs translating, which is somewhat compounded by the fact that it’s not simply about translating content, it’s about ensuring that all content is accessible to as many customers as possible, quickly, in a language they can instantly understand. So what can be done?
Use technology you can trust
Machine Translation (MT) is now forging its way into the mainstream and the innovators of the localization industry are grasping it with both hands to experience massive gains in productivity. The often-held perception that MT quality is not good enough is starting to change as we see increasing maturity in this area.
At SDL we have made great progress in MT quality, built on over 20 years of experience. We have moved far beyond the traditional one-size-fits-all approach, the results of which can often depend on the language pair and content type. Our innovative new SDL XMT technology has been designed to be flexible and adaptive, combining a series of powerful algorithms which allow for higher MT quality, particularly for challenging language pairs such as English to Asian languages. XMT also supports new content formats and informal languages such as texts and tweets and provides a platform for improved MT capabilities and customization. So XMT is really beginning to set new expectations for what is achievable with MT, and with our new addition to SDL LC you can set a new standard for your own MT quality by training a custom MT engine.
Boost productivity and minimize post-editing
So what is a custom MT engine? The good news is that you can start by using assets that you already have to get even more out of your existing content. Custom engines are trained using these assets and are merged with our latest high-quality XMT baseline engines. Users of MT find that it improves efficiency three-quarters of the time² and by training your own custom MT engine you’ll benefit from relevant MT output that reflects your own style and terminology, or that is suited to a particular domain or client. This will require less post-editing, in turn helping to increase overall productivity and content quality so you can experience similar gains in efficiency.
Easy to get started
It’s easy to get started in our simple yet powerful training environment, which is secure and cloud-based so you have no additional hardware or maintenance costs:
Simply upload your TMX files and these will automatically be cleaned and prepared to optimize the engine training for the best results. The wizard will guide you through the process:
Once complete, activate your engine and use it immediately within your translation workflows. We have found that intelligently trained MT engines translate industry-specific terminology 24% more accurately than untrained engines, so you can ensure that industry-specific terminology is translated accurately and consistently, regardless of user.
SDL LC Custom MT Engines can help equip you to be better prepared than ever before to take on more projects and meet increasingly tight deadlines, but without having to compromise on cost or quality. With a number of different packages available to suit your business requirements, you can reduce manual processes and embrace advances in MT technology to experience higher efficiency through reduced post-editing, across all content types. SDL LC Custom MT Engines are currently fully integrated with SDL Trados Studio 2015, with Studio 2014 integration coming soon.
1 – John Gantz and David Reinsel, The Digital Universe in 2020: Big Data, Bigger Digital Shadows and Biggest Growth in the Far East, IDC iView (December 2012)
2 -Based on SDL research.