My greatest nightmare with Studio's recognition of terms persists. I thought that "enable overlapping terms" option would help, but unfortunately it doesn't.
Once again the nature of the problem: Studio doesn't display two-word terms, whenever there is o one word term containing one of the elements of the two-word one. For example: I have just added term "OBCY ZAPACH" in Polish to the Termbase with the English translation "FOREIGN SMELL".
However, Studio refuses to recognize it in the text, because one of my other Termbases (but in fact it could be even the same termbase) contains term "ZAPACH", so Studio displays only this entry, completely ignoring "OBCY ZAPACH".
I can post a video if you want. It's extremely annoying and there's no way to rectify it.
Hi Nora, The video is here: https://youtu.be/s93z6cfjuKA As you can see, despite everything I did to have the terms displayed, I got nothing.
I'm using the demo version of Studio 2017, since the bug with non-display of digits in 2017 forces me to stick to my Studio 2015 license for projects with a lot of numbers.
UPDATE: After a lot of attempts I think I know where the problem lies. It seems that Studio is unable to recognize fuzzy termbase matches if the first element of a compound term has a slightly different form. Just one letter difference is enough for Studio to go completely blind to such terms.
So, OBCY ZAPACH is recognized because it's a 100% match ("OBCY ZAPACH" exists in the termbase). "OBCY ZAPACHU" is recognized (though in fact it's not grammatical in Polish), because the change in inflection is at the end of the whole term. But "OBCYM ZAPACHEM" is NOT recognized, simply because the inflection of the first word changes, which in Polish is crucial, because both nouns (ZAPACH) and adjectives (OBCY) decline grammatically.
Now, my great request to SDL: could you change it? It's really important for those whose languages are heavily inflected!
I was just going to say this after watching your video, but I see that you've updated the post with your findings. Changing "obcy" to "obce" in the termbase entry results in proper recognition of the overlapping terms, but I also find (from a quick test with your example) that "zapach/zapachy" may actually be the issue here. For me, the overlapping terms aren't recognized only when both zapach and zapachy are in the termbase. If only obcy zapachy and zapach are in the termbase, but not zapachy by itself, then both terms are listed:
Yes, that's true. Unfortunately, this is a giant problem for inflected languages because inflections of compound nouns in Polish and many other languages influences both components of compound nouns. I wish SDL would do something about it, because it makes termbases only half-usable at the moment.
ANOTHER UPDATE: I'm at a complete loss here. It seems like your findings are true. Anyway, something should be done about it, because a blind CAT is a sick CAT :(
TO SUM UP, in my case it looks like this: we have three terms - "OBCY ZAPACH", "OBCY" and "ZAPACH".
In sentence: "Poczuł obcy zapach" - ALL THREE are recognized.
In sentence: "Poczuł obce zapachy" - only "OBCY and "ZAPACH" are recognized (Studio is blind to compound "OBCY ZAPACH")
In sentence: "Poczuł obcy zapas" - only "OBCY" is recognized.
In sentence: "Poczuł obce zapach" (ungrammatical) - only "OBCY and ZAPACH" are recognized separately (Studio is blind to compound "OBCY ZAPACH").
I must say that I've got fed up with all those problems in Studio. There are too many issues I must deal with. All of the sudden, MemoQ seems like a less troublesome option.
I wonder if it has something to do with the "fuzziness" of the match.
Wojciech Matyszkiewicz said:For the time being I had to return to good old Wordfast Classic.
Oh no ;-) More seriously though, it's good to have this kind of detailed information so the team can address this problem. Hopefully we can help put Wordfast to bed ;-)