Menu
▾
▴
#1306 OmegaT treats as repetitions segments that are not identical
How come OmegaT thinks the following two segments are identical?
- <x0> has received the highest number of emails (<x1>).</x1></x0>
- <x0> has the highest number of recipients (<x1>).</x1></x0>
Something weird happens with stemming and leads to such problems. It also makes OmegaT insert automatically TM matches that are not 100%, although I have set it to only insert 100% matches.
By the way, I am using 6.0.1, but I could not find it in the list of "Milestones".
We need a sample project to reproduce the issue.
Can't share the project, sorry. It should be easy to reproduce, though.
Right now, my source segment is:
Books
And the translation memory has:
Bookings
The translation of this TM hit is inserted in the editor as a 100% match.
The lack of match percentages in the editor pane, along with the lack of a confirmation mechanism for segments, make this behaviour highly dangerous, and decreases my confidence in OmegaT for mission-critical work.
We need a sample project. We don’t need you to share the original.
Just create a source file with two different segments that are considered identical in OmegaT. Put that is a project with a corresponding 1 segment TMX and check that you reproduce the issue before sending it to us.
We’ve had OmegaT used in mission critical work for more than 20 years by thousands of professional translators who depend on reliable tools to pay their bills and we’re pretty reactive on bugs that we can reproduce.
When OmegaT inserts a match above the set threshold, it also inserts a marker that indicates that the match was automatically inserted.
To confirm that the insertion is valid, the user either manually removes the marker, or uses the "insert match" function to manually insert it.
I’m not sure I understand what you mean by "the lack of match percentages in the editor pane". Which percentage should be displayed ?
Here it is. Just two segments. Once you translate "Bookings", it auto-completes it as a 100% match for "Books".
Please, make sure that the project opens properly on your side.
Let's just leave it here. The app does stemming for TM matches, which is clearly incorrect. If you don't think this is a bug, that's fine with me.
Last edit: Epameinondas Soufleros 2025-10-09
If you have specific requests like:
feel free to create a RFE.
The project that you sent behaves properly and OmegaT does not consider the two segments as equivalent : If you put the first translated segment in a TMX inside /tm/auto OmegaT will not automatically translate the second segment.