Custom Neural Machine Translation Engines

Custom Neural Machine Translation Engines (CNMTEs) are built specifically for clients that have a large volume of translations in a very specific domain (i.e., content vertical). These custom engines, when properly built and trained, can be a significant asset to any organization that needs to translate a large volume of highly specialized content. Trusted Translations is on the cutting edge of this technology and is one of the world leaders in providing high-quality translations using Custom Neural Machine Translation Engines. By leveraging our technology, companies can achieve high-quality translations of vast amounts of information, all at a fraction of the cost and time involved in a typical human translation process.

The following are some of the typical attributes and objectives of translation projects that may be good candidates for our Custom Neural Machine Translation offering:

  • Large volume of content
  • The need for fast turnaround times
  • The highest ROI
  • Highly specialized content

Projects that once were deemed impossible or cost-prohibitive using a typical human translation process are now possible through the implementation of a Custom Neural Machine Translation solution. Trusted Translations has an IT team dedicated to implementing these client-specific solutions, and they are available for a free consultation and evaluation to help you determine whether the whole volume of your translation needs, or perhaps certain particular scopes, can benefit from this technology.

Just keep in mind that a CNMTE is not an impossible dream. Improved technologies, more accessible platforms and services, and smaller volumes of corpora are all factors that contribute to bringing this powerful solution within your reach.

Custom Machine Translation and Human Post-Editing Workflow

As you can see in the illustration of a sample Custom Machine Translation Engine workflow, the process can be integrated directly to the client’s Content Management System (CMS). This facilitates the process and eventually leads to faster turnaround times and seamless deliveries. In this case, the Translation Management System merges a Customized Machine Translation and Human Post-Editing into a single environment. Furthermore, corrections applied by the post-editor on the machine translation output not only go back to the client’s CMS for final publishing, but are also fed back into the Custom Neural MT Engine for machine learning and incremental retraining, rendering the engine more “intelligent.” Hence, quality improves significantly over time. This is especially true for projects involving high volumes of content in the same domain (industry vertical).

Integrating specific terminology on top, such as domain- or client-specific glossaries, can contribute to increased accuracy and consistency. Such integrations involve placing terms within the proper context, with terms always expressed in their most accurate form based on prototypical inflectional categories such as number, tense, person, case, gender, etc. This correctly applied terminology ends up taking part of that retraining process.

Developing a Custom Neural Machine Translation Engine (CNMTE) will incrementally improve the quality of the output over time, and it will build an asset your competitors may not be able to acquire. CNMTE leverages state-of-the-art technology to improve automated content through the use of expert human post-editors.

Please contact us today, and we will schedule a free evaluation to help you determine whether this solution is right for you and your organization.