Configuring machine translation adapters

WorldServer supports machine translation as an alternative source for fuzzy matches. During the leverage process, WorldServer uses both translation memories (TMs) and machine translation (MT) to produce fuzzy matches. The leverage rules are the same for matches resulting both from TMs and from MT: SP/ICE matches are preferred, then 100% matches, then fuzzy matches.

In the absence of 100% matches (or better), the score of the best fuzzy match is recorded and used for scoping purposes. If the score of the best fuzzy match resulting from TMs and from MT is equal, the TM match is preferred.
You can find the following machine translation adapters in the WorldServer distribution kit:
Machine translation adaptersLocation
SYSTRAN, Uppercase Pseudotranslator, and Lowercase PseudotranslatorIn the SDK, under samples\dist\mt_samples.zip
Language Weaver EdgeIn the distribution kit, under integrations\Language Weavers\Language Weaver Edge
Language Weaver CloudIn the distribution kit, under integrations\Language Weavers\Language Weaver Cloud
After you configure machine translation on your environment, you need specify which MT configuration WorldServer should use when calculating leverage. You can do this on the following pages:
  • On the configuration page of each project type (Management > Business Rule Linkage > Project Type).
  • On the Management > Asset Interface System > View and Change Properties page for each target folder.
When you open a version of an asset for the first time, the leverage process is triggered and uses the machine translation adapter you configured as an alternative source of fuzzy matches. In general, WorldServer uses the adapter specified in the AIS properties. However, for projects that have a project type, the adapter specified in the project type is used instead of the one specified in the AIS properties.