Documentation Center

Support for the new Language Weaver plugin with MTQE capabilities

The legacy Language Weaver Cloud and Language Weaver Edge apps have now merged into the more efficient Language Weaver Provider app .

The Language Weaver Provider offers secure, highly adaptable, and scalable AI-powered machine translation. Its built-in Machine Translation Quality Estimation (MTQE) feature assesses the quality and reliability of machine translation output coming from Language Weaver Provider. This enables you to determine when human editing is needed, and evaluate the suitability of your data for machine translation.

With CU11 release, Trados GroupShare now integrates with this new plugin, enabling you to leverage Language Weaver's machine translation capabilities through project templates created with Trados Studio.

This means that Trados GroupShare projects that include Language Weaver in the project template will benefit from this translation provider's capabilities.

However, MTQE assessments can currently only be displayed in the Studio's Translation Editor and cannot be viewed in the Online Editor. Also, the Online Editor cannot display MT results in the lookups panel.

To use the Language Weaver Provider plugin with Trados GroupShare projects, keep in mind the following important aspects when setting up app in Trados Studio.

  • By default, Trados GroupShare uses Language Weaver's European Language Weaver server, regardless of the initial selection made during the template creation process in Trados Studio. To use the United States server instead, open the ExecutionService.exe.config file available under C:\Program Files (x86)\SDL\SDLServer\Application and change this value of the LanguageWeaverCloudURI setting to:https://us.api.languageweaver.com
  • When uploading Pre-/Post-Lookup and Find/Replace files, make sure to include the files containing the replacement rules from a network drive accessible on the Trados GroupShare server hosting your project. These .XML files allow you to fine-tune your Neural Machine Translation (NMT) output in conjunction with NMT dictionaries
  • When choosing a machine translation model for the project templates, if you want to include one that also offers quality estimations, choose a model whose name ends in "QE"

  • When creating the project template for Trados GroupShare, including the Apply Language Weaver Metadata batch task is not mandatory. Unlike Trados Studio, GroupShare automatically applies metadata, including quality estimations, as part of the Pre-translate batch task. If the Apply Language Weaver Metadata task is unintentionally included in the Studio project template, Trados GroupShare will simply disregard it.

Known limitations

  • The Language Weaver Provider app is unable to generate results for any project languages added to the project after uploading the project template.
  • Trados GroupShare cannot process project templates that use the GroupShare Analyze and Translate batch task when the Language Weaver Provider app or any other Machine Translation app is also specified in the template.