As explained in this section of the RWS Documentation Center, WorldServer relaxes the requirement for tag parity in translation units by default. In other words, some tags may be present in the target text without necessarily being in the source text and vice versa. Trados Studio creates return packages for WorldServer without enforcing tag parity. Note the following:
studio_missing_formatting_penalty studio_different_formatting_penalty that is set to 0.01 value per default play a role in enforcing the relaxed tag penalty? This is the relevant section of tm.properties default settings: ### Studio Penalties ## Apply only in Studio Scoping Mode # The Missing Formatting Penalty is triggered if the document source segment does # not contain any tags, but the translation unit's source segment does. In this # case, all aligned tags are dropped from the translation proposal and the penalty will be # set. Unless there are target-only tags in the translation unit which cannot be # resolved against the current filter's quick tag list, the different formatting # penalty will not be triggered in addition. #studio_missing_formatting_penalty=0.01 # The different formatting penalty is applied if the translation # memory and the document source have different formatting, # or if additional formatting in the translation memory target # segment cannot be fully resolved by using filter-specific # quicktag information. #studio_different_formatting_penalty=0.01 |
With the current default Studio penalties set in WorldServer environments: studio_missing_formatting_penalty=0.01 studio_different_formatting_penalty=0.01 when a Return package is imported into WorldServer containing segments where additional Studio tags have been added to target segments that are not present in the source segment, or the target segment has different tags (i.e. with different Tag IDs) that the source segment, or missing tags in the target segment while present in the source segment, the Return package will still be imported without error. Examples of such segments in Studio (the Tag ID View enabled): These segments will be stored in the WorldServer TM as follows: The TM leverage from such TM entries will have a penalty of 0.01%. For instance, if a file has been previously translated and is submitted as a new project, the TM leverage for an identical source segment where the TM entry contains - different tags (i.e. with different Tag IDs) in the target segment versus source segment - or missing tags in the target TM entry while present in the source segment - or the target TM entry has additional tags/placeholders than in the source TM entry will result in a 99% fuzzy match instead of a 100% match: However, if you remove the 2 penalties from tm.properties by setting them to 0 studio_missing_formatting_penalty=0.00 studio_different_formatting_penalty=0.00 and restart Idiom Service to apply these changes, the TM leverage will be 100% Match and WorldServer will also "repair" the target segment by removing the additional or non-matching tags. Remember: such TM entries will never be ICE Matches. With this change, if your process is set not to review 100% matches, your target file might have different formatting than the source file. So to take advantage of the high TM leverage despite tag discrepancies, you should implement a review of 100% matches to ensure that any tag discrepancy is addressed. |