Details
-
Bug
-
Resolution: Fixed
-
None
-
High
-
In the communication with OpenTM2 the used tags are modified to improve found matches.
-
Empty show more show less
Description
problem
Assume a TMX containing a lot of single tags in the form <it type="struct"/>.
If we try to find matches we never get 100% matches, we always got a penalty in the match rate.
For reproduction assume the following 5 Segments:
Segment 0: Anzahl Tests: <x id="1"/>
Segment 1: Anzahl Tests: <it pos="end"></mq:rxt displaytext="Tag2" val="@XXX_UPD$1"></it>
Segment 2: Anzahl Tests: <ph></mq:rxt displaytext="Tag2" val="@XXX_UPD$1"></ph>
Segment 3: Anzahl Tests: <it type="struct" />
Segment 4: Anzahl Tests: <it id=\"2\" \/>
If we request this TM with the segment "Anzahl Tests: <tag>" and the tag in different variants, we got different results.
Tag Variants requested | Resulted Segments (id and matchrate) |
---|---|
<it pos="end"></mq:rxt displaytext="Tag2" val="@MNCS_UPD$1"></it> | 1: 100 3: 13 0: 13 |
<g id="1"></g> <it type="struct"/> <x mid="1"/> |
0: 100 3: 100 4: 100 2: 13 1: 13 |
<x id="1"/> → This is the form we currently use | 0: 100 2: 13 1: 13 |
So with changing the single tags from <x id="1"/> to <x mid="1"/> is little effort but improves the found segments a lot.
Since this is a problem in OpenTM2 so far, a issue is created there too.
Since we can fix this on our side with little effort, we implement this fix.
Attachments
Issue Links
- relates to
-
TRANSLATE-2338 Use ph tag in OpenTM2 to represent line-breaks
- Done
-
TRANSLATE-1481 Improve tag handling with matches coming from OpenTM2
- Done