With XTM, harnessing extra value from your language assets : The new feature of XTM and SYSTRAN

“By using XTM and SYSTRAN, we’ve been able to find extra value in our language assets. We had a phenomenal setup with our content management system feeding automatically into and out of XTM. Adding SYSTRAN MT to the process made it even better.” Ronald Egle​, Content Systems Administrator Ariel Corporation

Generic vs. Customized Neural Machine Translation

achine Translation (MT) has become a strategic asset for enterprises that need to translate large volumes of content quickly and consistently. As professionals demand higher translation quality and productivity, the importance of customization has grown significantly. To meet diverse use cases across a company, customization must be fast, simple, and highly effective.Evolution of Machine Translation Technology

• Neural MT (NMT) : Delivered a breakthrough in fluency and accuracy. Yet, as quality improved, professional users demanded even more tailored translations, making custom NMT the new standard.
• Rule-Based MT (RBMT) : Early systems relied on manually crafted rules. Customization was essential to achieve usable output.
• Statistical MT (SMT) : Marked a big leap in quality by leveraging large data sets. However, customization was still limited.

How Fuzzy Matches Enhance NMT

XTM’s AI-enhanced Translation Memory (TM) feature boosts MT quality by sending fuzzy matches to the neural engine as context.

• What are fuzzy matches?
  They are similar — but not identical — sentences with a high match percentage from your translation memory.
• How it helps:
  Validated TM entries act as “hints” for the MT system, helping it generate better translations on the fly and reducing the need for human corrections.

This synergy between translation memories and neural MT optimizes results in real time, leading to faster, more consistent output.

Why Validated Translation Memories Matter

A company’s translation memories are a strategic asset. With the AI-enhanced TM feature, previously approved translations guide the MT engine to deliver higher-quality results. This approach minimizes repetitive corrections and ensures brand and terminology consistency.

Measurable Impact

By combining neural fuzzy matches with domain specialization, companies can achieve up to +22 BLEU points improvement (18 from neural fuzzy + 4 from specialization). This directly translates to faster time-to-market and reduced post-editing costs.

About XTM

Founded in 2002, XTM International identified a gap in enterprise-class translation technology and created XTM Cloud TMS — a powerful, browser-based Translation Management System integrated with a CAT tool. Its open standards, scalability, and security make it ideal for enterprises needing to localize large, fast-changing content.