Downloading MT Translation Kits

Related Information:

Training

Troubleshooting and Advanced Tips

FAQs

MT segments can be pre-populated in the bilingual files for RTF, TTX, and XLIFF file types using checkbox on download page. Users now can get MT word count information on the Detailed Word Counts web page.

Downloading kits for CATs

• MT is applied to a project in GS via API to the MT system.
• Typically, 75% matches and higher go to the TM for leverage, while everything below that threshold goes to the MT system. Note that MT segments will be identified by match value of 60%
• For each download option in GS, the MT segments are stored in the TMX of the downloaded kits
  
  
  
  
• When MT has been activated for a specific workflow, MT will normally replace the new words segments.
• MT matches can be downloaded into a separate TMX, when you tick the checkbox option MT matches into separate TM file. This is the recommended option.
• Caution: leaving the option unchecked will result in MT in the project TM.
  
  
  
  
  
  
  
  

MT segments are stored in the TMs of the downloaded kits

• Caution: MT segments are stored with 'normal' matches in the project TMX of the downloaded kits when Include MT matches as separate downloaded file is unchecked.
• Include MT matches as separate downloaded file checked is the recommended setting
• In the TMX, MT segments are identified with the creation ID “MT!”
  
  
  
  

Importing TMs into Master TM

• Caution: When importing content from the downloaded TMs into your master TM, please only import the Fuzzy match TMX file and not the MT TMX file.
• Because the MT matches are already populated in the files (“Populate MT Target Segments” checked) you don’t have to import the MT TMX into your master TM.
• The point of downloading them as a separate MT TMX file is that this is to be discarded when importing content into your master TM.

OmegaT Download Option

• When MT has been activated for a specific workflow, a separate MT TMX is automatically provided inside the penalty-26 folder
• MT segments will show as 74% matches in the OmegaT Editor
  
  
  
  

Bilingual RTF Download Option

• Tick Include MT matches as separate downloaded file. This is the recommended option, otherwise MT will populate the project TMX
• Populate MT Target Segments option must be turned on in order for MT segments to be pre-translated in the bilingual RTF file (MT match value 60%)
  
  
  
  
• 'Resources' are marked as MT or TM depending on the source
• MT match value is '60%' (unless project instructions say otherwise)
  
  File:07 offline rtf percent 60.png

  912px
  
  

Trados 2007 TTX Download Option

• Tick Include MT matches as separate downloaded file. This is the recommended option otherwise MT will populate the project TMX
• Populate MT Target Segments option must be turned on in order for MT segments to be pre-translated in the bilingual TTX file (MT match value 60%)
  
  
  
  
  
  File:08 offline ttx percent.png

  523px
  
  

Xliff 1.2 or 2.0 Download Option for Studio / memoQ

• Tick Include MT matches as separate downloaded file. This is the recommended option otherwise MT will populate the project TMX
• Populate MT Target Segments option must be turned on in order for MT segments to be pre-translated in the bilingual XLIFF file
  
  
  
  

Importing TMs into Studio

• Import the separate MT TMX into Studio TM
• In the Penalty section, set it to the maximum allowed by Studio 25%.
• When project files are opened, all MT matches will be penalized and display as 75% matches.
  
  
  
  
  
  
  
  

How to Test the MT Workflow

It is important to test the TMS-MT-CAT tool roundtrip before launching your first MT project:

• Make sure to follow the download recommendations for your preferred CAT tool
• Confirm that the MT segments can be identified in the TM and the Editors you are working with
  • TIP: Check the pop-up Editor in GS to make sure MT is activated
• Orange segments in the editor are MT. Use the filter: If you see orange MT segments it confirms that the MT connector in GS is working.
• Review the setting in your CAT tool if there is MT in the GS Editor, but you cannot see MT in your CAT tool.
  
  
  
  

How to Read the GS Word Count Report

• Unless the word count report identifies MT as a category, in most cases the No Matches represent MT words
• In GS, the No Matches or the 50-74% category represent MT words
  
  
  
  
• Although the official word counts come from the GS logs, many teams run their internal word count reports in Trados or Studio
• In the Trados logs, the category 50-74% represent the MT segments when the MT penalty has been set to 26% in Trados Workbench
  
  
  
  

xxxxxxxxxxxxxxImporting TMs into Trados Workbench (RTF, TTX)

• Trados Workbench recognizes the MT! label in the project TMX
• When importing the project TMX or the separate MT TMX into Trados Workbench, adjust the MT penalty to 26% (consistent with GS confidence score of 74%) to avoid mixing low Fuzzies and MT