• State Suggested Answer
  • Replies 6 replies
  • Answers 1 answer
  • Subscribers 432 subscribers
  • Views 3646 views
  • Users 0 members are here
Options
Related

Sentenced-based TM re-segmentation of (imported cell-based) aligned/bilingual content (Excel/XLIFF/SDLXLIFF)

Manuel Hörmann

Hi,

I have done an alignment (German/English) via the bilingual excel file type and the complete Excel cells (more than one-sentence, the same happens with bilingual (SDL)XLIFF files) were saved as TM segments in the Studio TM (see attachment). The TM segmentation, however, is set to full stop rule (sentence-based).

I could have copied them into 2 excel files and perform a traditional alignment which would have worked, too, I know. But I wanted to show it using just 1 file. The result – referring to the TM segmentation - strikes me as odd :/. As the (customer-related) data is confidential, I do not post it or any screenshots of it here.

If I import a new excel sheet for translation using the same TM which will then be segmented according to the full stop rule, I can only use upLIFT matching or concordance search, but not fuzzy matching (< 70%).

As far as I know, there is no easy way to re-segment TMs in Studio using the features integrated (e.g. TM maintenance)? Am I right? I also read the following related article that you cannot change segmentation rules for the bilingual excel file type in Studio:

https://community.sdl.com/solutions/language/translationproductivity/f/90/t/8575

I wondered whether you had experience with tools such as Olifant or even a useful script for re-segmenting a Studio TM or exported TMX file? Is there an app for this? I could not find any on the AppStore.

Thanks a lot!

Best regards,
Manuel
Rate translation
×
Suggest better translation
×
Moderator UI
×
  • Get AI Suggestion
Parents Reply Children
  • 0 in reply to Paul

    Thanks, Paul! I'll definitely check the SDL Convert option regarding my use case. I did not mean to confuse you or anybody. However, I had a multilingual file (6 languages contained in separate columns in one Excel file) which I thought would be a brilliant use case for the bilingual excel file type without much manual preparation. I even tried it with a TMX creation using the Glossary Converter that produced the same unsatisfactory (cell-based) result regarding segmentation. In case of a traditional alignment, I need to split up a file into individual files for alignment which is not required when you use the bilingual excel file where everything is already contained and you just have to point to the respective columns. The downside, however, is the (non-configurable; cell-based) segmentation that you have to fix afterwards. My idea would be that even if allow cell-based segmentation for a sentenced-based TM, that you might fix the segmentation afterwards while re-importing (via SDLXLIFF, TMX etc.). Other CAT tools already provide such an option while importing (Import tranlsation unit with sentence segementation):

    Otherwise, there will be no or significantly fewer matches (given the >=70 percent fuzzy match settting) and customers starting with little legacy data might not fully benefit from upLIFT (especially, TU fragments) and TM reuse.

    Regards,
    Manuel