Knowledgebase: Product > memoQ
Post-translation analysis pitfalls
Posted by P├ęter Botta on 19 January 2013 02:01 PM

Title: Post-translation analysis pitfalls

Description:

For the basic mechanism of how post-translation analysis works, see the memoQ help topic:

http://kilgray.com/memoq/2013/help-en/index.html?post_translation_analysis.html

Post-translation statistics counts confirmed segments, and it only counts segments that the specific user, assigned to the document, confirmed. That is, if some of the segments are confirmed by a project manager or administrator, not assigned to the document, those segments will be invisible to post-translation analysis.

Also, a word of caution, although post-translation statistics is a useful tool, there are a few things you need to know: 

Accuracy and known issues of post-translation analysis

There are ways to confuse or "fool" post-translation statistics.

-Reliance on "Automatic lookup and insertion" before version 5.0. Before version 5.0, the calculation relied on "Automatic lookup and insertion". "Automatic lookup and insertion" is basically an on demand pre-translation: it pre-translates the current segment and fills the match reate value that is seen next to it. Post-translation statistics uses those values to figure out what kind of matches the user actually got during translation. If a translator turns off "Automatic lookup and insertion", post-translation analysis will give no useful result. Or if a translator plays cleverly, like turning it on and off, she can earn more than they should, some of the leverage she got gets hidden.  "Automatic lookup and insertion" itself only works on segments that are empty on the target side (more specifically, in "Not started" status). If you (or a translator) pre-process the documents in some way so that they are not all empty when it is translated, the translator can gain money, leverage might be hidden.

-Since version 5.0, TM hits inserted manually by the translator are also calculated, making PTA much more useful. Whatever match rate percentage value you see next to the segment is taken into account in the calulcation.

-Translators using their own TMs. If a translator adds her own translation memory, and gets matches from it, she can lose money, which can be unjust.

-Sliced documents don't work with post-translation analysis in any version before 6.2.19. If you worked with sliced documents in an earlier version, you need to reconfirm the segments after updating to 6.2.19.


(0 vote(s))
This article was helpful
This article was not helpful

Comments (0)
Help Desk Software by Kayako support.memoq.com/index.php?