Modules and settings

Several modules present in a pipeline flow are configurable for certain attributes. This allows you to customize your translation flow and manage a balance between speed, quality and cost, empowering the LangOps user to remain on top of any results.

Other modules have no possible configuration, being merely informative even though they may be mandatory for any pipeline.

To configure a module's settings click on it and see a box with available options pop up. Find a description to each module's configurable settings below.

Anonymisation

Anonymisation occurs by default in any translation. Clicking the module will display the type of name categories we detect and anonymize.

Glossary (configurable)

Selecting a glossary will apply it to all pipelines in the translation group. Any included terms will be translated according to the Glossary composition.

Translation memories

The server ID that stores your translation memories is displayed here. TMs are managed under Linguistic Resources.

Machine translation

Details on the translation model used by the pipeline can be found here.

Quality estimation

More details coming soon.

Decision

Adding this model allows you to have some translations skip human edition, if included in the workflow, based on a quality score threshold. As the levels increase, more translations will be skipped.

Human translation

You can add or change editor instructions. These will display to human editors when working on a translation. These instructions are based on the default for the Pipeline group and can be further customized per pipeline here.

It's also possible to establish a timeout value for human edition. If you prefer to wait for a human to check the MT translation indefinitely, leave the checkbox ticket. If you want to set a time limit which, if hit, delivers the MT output directly, uncheck it and set the intended duration of the limit.