SDL announced XMT, a core component of the language pillar within SDL’s Customer Experience Cloud.
Developed by the SDL Language Research Group, XMT is a rewrite of machine translation technology developed to address the limitations of previous generations of machine translation (MT).
The platform allows different translation algorithms to be used for different language pairs. It also allows for the development and deployment of new MT capabilities, including SDL Language Learning. This will bring artificial intelligence to machine translation by empowering the engines to learn and apply individual user language preferences.
"The biggest multilingual challenge for today’s international businesses is the dramatic shift in the way that organizations communicate and engage with their audiences," Kirsty Waller, vice president of marketing for language solutions at SDL, told eWEEK. "Customer communities now span cultures and countries. Customer segments are no longer defined by their geographic borders--rather they are defined as groups of customers who have specific likes, dislikes, desires and needs. This shift leads to the need for all communication to be delivered in multiple languages rather than defining what is translated by country."
In addition, translation engines learn, adapt and improve translated output based on feedback from human reviewers.
"International big data, particularly social big data, holds a wealth of opportunity for understanding customers and building better experiences on that insight," Waller said. "However, without the ability to analyze and interpret data from around the world organizations are not able to truly harness the power of this information. The volume and velocity of big data requires an automated approach to translation that fully integrates with data collection and analytics software."
XMT will be made available through SDL’s suite of machine translation solutions including Language Cloud and BeGlobal, and will provide support for more than 100 language pairs and five industry-specific vertical engines.
The XMT engines can be further trained on an organization’s specific vocabulary and terminology to produce a customized solution. As part of the overall translation ecosystem, XMT engines can also be coupled with human post-editing to further improve translation quality and reduce time-to-market for translated content.
"As global communities continue to build and evolve, the need for real-time translation is going to grow. There are not enough human translators in the world to deal with the demand," Waller explained. "Organizations need to adopt new ways of communicating and close the gap on what machine versus humans can translate to accelerate the process and improve efficiencies. Allowing humans to focus on content that requires the human touch, the advent of artificial intelligence and machine translation will bring the rest over the goal line, ultimately closing this gap."