Analyse shows some 95-99% matches in ttx, but in TagEditor I can find only 100% matches Thread poster: Erik Freitag
|
Erik Freitag Germany Local time: 06:13 Member (2006) Dutch to German + ...
Dear colleagues, I guess I am misunderstanding something, but I can't find out what's happening. So here goes: I have translated a TTX file (in TagEditor, obviously). I have translated all segments, when I open the first segment and hit "translate to next fuzzy match" (or whatever this function is called in English, ALT+NUM*), TagEditor runs all through to the end of the text. So far, so good. When I analyse this file with the same TM in TWB, the analyse shows 4 s... See more Dear colleagues, I guess I am misunderstanding something, but I can't find out what's happening. So here goes: I have translated a TTX file (in TagEditor, obviously). I have translated all segments, when I open the first segment and hit "translate to next fuzzy match" (or whatever this function is called in English, ALT+NUM*), TagEditor runs all through to the end of the text. So far, so good. When I analyse this file with the same TM in TWB, the analyse shows 4 segments as 95%-99% fuzzy matches! What's happening? I just can't find them in TagEditor! Any help appreciated! Kind regards, Erik ▲ Collapse | | |
Have you checked your Minimum match value? | Apr 10, 2008 |
If you set the Minimum match value to 100% on the tab Options / Translation Memory Options / General, then it should stop at the matches below 100%. As far as I know, the Trados fuzzy match analysis has never been 100% compatible with itself, but that's not the only reason why it is a secret algorithm.. | | |
Erik Freitag Germany Local time: 06:13 Member (2006) Dutch to German + ... TOPIC STARTER Changed settings - same behavious | Apr 10, 2008 |
Dear Harry, Thanks for your tip, I've tried to change the setting you mentioned, but the behaviour doesn't change. Strange. Any other ideas? Kind regards, Erik | | |
Accept the existence of miracles.. | Apr 10, 2008 |
efreitag wrote: Any other ideas? Be happy! Wouldn't it be worser the other way round, if the analysis would show only 100% matches, but the Translate to Fuzzy option would not find all of them? Generally, I would not expect any strict logic in a fuzzy logic environment. | |
|
|
Erik Freitag Germany Local time: 06:13 Member (2006) Dutch to German + ... TOPIC STARTER Ok, I'll stop worrying | Apr 10, 2008 |
Harry Bornemann wrote: Generally, I would not expect any strict logic in a fuzzy logic environment. That's probably true, but it's just one single fuzzy logic here, giving two different results. I am pretty sure I haven't missed anything in the translation, so no reals problem so far. I was just wondering what's happening here... Thanks anyway! Groetjes, Erik | | |
Heinrich Pesch Finland Local time: 07:13 Member (2003) Finnish to German + ... Export the less than 99 % matches | Apr 11, 2008 |
Then you'll see what's happening. Cheers Heinrich | | |
Erik Freitag Germany Local time: 06:13 Member (2006) Dutch to German + ... TOPIC STARTER
Dear Heinrich, Thanks, that was a good idea. I did that. The export file showed the four segments. These contain a translation in my document, while in the export file, the target is identical with the source. I can't explain that, either, but at least I know I haven't overlooked anything in my translation. Any insights about this issue are still appreciated (though just out of interest)! Kind regards, Erik | | |
Minimum match value serves a different purpose | Apr 11, 2008 |
Harry Bornemann wrote: If you set the Minimum match value to 100% on the tab Options / Translation Memory Options / General, then it should stop at the matches below 100%. Sorry, Harry, but that setting has no bearing whatsoever on Erik's question or on the "Translate to Next Fuzzy" feature. I'm just saying this to prevent future readers of this thread from getting the wrong picture. This setting is used to adjust the threshold below which a "Fuzzy Match" is no longer considered fuzzy, but a "No Match" instead. Therefore, changing this setting affects the number of fuzzy matches returned when analyzing or translating a file. Nothing more and nothing less. As far as I know, the Trados fuzzy match analysis has never been 100% compatible with itself, but that's not the only reason why it is a secret algorithm.. My observations over the years lead me to believe that the segmentation algorithm in Workbench follows different rules than the algorithms in the Word macros and in TagEditor. In my experience, the inaccuracy is often related to segments containing numbers and/or placeables. The analysis algorithm is probably performance-optimized at the expense of some accuracy when it comes to such segments. All the best, Benjamin
[Edited at 2008-04-11 12:30] | |
|
|
Tudor Soiman Local time: 07:13 Member (2005) English to Romanian + ...
Save a copy of the translated file and clean it up. Analyse the original. You'll have 100% matches.
[Edited at 2008-04-11 13:16] | | |