• State Suggested Answer
  • Replies 3 replies
  • Answers 1 answer
  • Subscribers 433 subscribers
  • Views 3088 views
  • Users 0 members are here
Options
Related

Perfect Match is not reliable

Bettina Weinmann

Hello,

I came across with a strange problem concerning Perfect Matches and I hope that somebody can help me:

We have to translate a manual into English and Japanese which was translated by us two times: October 2016 version A,  May 2017 version B.

Now version C has to be translated. When preparing the Trados Studio project I used the files from version B for Perfect Matches. The result was different for English and for Japanese. For English, 90% was pretranslated and for Japanese only 50%. My first thought was, that the Japanese analysis differs because we translate Japanese with English as source and as English was also a translation there might be differences due to formal issues, terminology changes etc. However, I started a sample project using the source files from version B and PM files also from version B and I thought that the result would be that no segments have to be retranslated. But I was suprised to see that only 50% was pretranslated using the files from exactly the same project! So I believe that PM only works once! Because when we translated version B we used the files from version A for Perfect Matches. Now for version C the matches from version B remained context matches (I compared the files). Did anybody made the same experience and knows a solution? Or is it a bug in Trados Studio? Is there a problem because of Trados Studio 2017? Our files in October have been created using Trados Studio 2015, now we are using 2017.

I'm looking forward to hearing from you,

Bettina
Rate translation
×
Suggest better translation
×
Moderator UI
×
  • Get AI Suggestion
  • 0

    Hello Bettina,
    I wonder why you didn't create a TM with your old xliffs and do then a pretranslate against this specialised TM? My experience is, that you get the best solution importing the xliffs without formating.
    I presume, your result would be a lot better using the pretranslate procedure in the normal configuration with 70 % fuzzy match in the TM.
    Try it out....
    PerfectMatch is perfect for reparing files, but in my opinion not really for mpreparing new jobs.

    Best regards
    Martha Ebermann
  • 0 in reply to Martha Ebermann
    Hello Martha,

    thank you for your reply. My problem is, that the client persists on PM, because the price is lower and they can be sure that nothing from the old versions has been changed.

    Best regards,
    Bettina
  • 0 in reply to Bettina Weinmann
    Hello Bettina,
    ok, I understand your point.
    Le problem with PerfectMatch is the following: if the old xliff used for Perfect Match has a similar content, but is based on a different file, a different software version or a different file type you will not get out the maximum of your existing translations, even if they match. When reparing something, it works perfectly on the same file where you can then easily repare broken segments or other things preventing you from getting a usable target file. But if you have the same file in a different Word version for instance, the tag differences will prevent you from getting out all your existing translations. If you want to get out the maximum of the existing translations, in my experience, it is better to “strip” the old xliffs from formatting by importing them in a TM in the plain text option and set the fuzzy value for pretranslate at 70 %. You will get a maximum of text and you would just have to replace the tags.
    If you have prepared your files in a smart way (for instance by locking segments containing only numbers, names or parts not to be translated) it will work better and faster than the PerfectMatch way.
    Is this explanation useful for you?
    Best regards
    Martha Ebermann