The
Calculate Repetitions Automatic Action scans through all of the tasks belonging to project (based on language) to find any segments that occur more than once.
Calculate Repetitions is a project scope automatic action, meaning that all tasks for a particular target locale must first reach this step before the action runs. After segmentation is complete,
WorldServer runs
Calculate Repetitions for all the assets in that language and records the repeated segment count in the Scoping information.
Note: if new Tasks/Files are added to a project
after the scoping report has been created and the repetitions have been calculated, the
Calculate Repetitions Step will be applied only to the new tasks since the original tasks have already passed the step, as explained in this article:
WorldServer - after adding a new file to my project, the scoping report does not update the Repetitions count correctly
Here are more technical details about how repetitions are calculated:
The Studio-aligned scoping mode ensures that
WorldServer handles repetitions in a way that reflects
STrados Studio repetition analysis.
Markup tags of the same type (singleton, opening, closing) appearing at the same location are considered equivalent for repetition analysis, as are segments which would be marked as 100% after repairs (auto substitutions).
WorldServer can identify all the duplicated segments (excluding all translated segments) within an asset and across the project. The following points describe the process:
- Segmentation and leverage occurs for each asset. The leverage process determines whether there are 100% or ICE matches available for each segment. If so, they will be applied and these segments will be attributed with a 100% score (and an ICE status if appropriate). If there are no ICE or 100% matches, then the best fuzzy score for each segment is recorded (0–99%) for scoping purposes.
- The repetition calculation process must be triggered. This must happen after segmentation and leveraging, and does not happen by default. The repetition counter goes through each asset and looks for duplicates among all segments that do not have a 100% score (which is why it is stated that only fuzzy matched segments are considered for duplication.) The first occurrence is designated as the repeated segment. Occurrences after the repeated segment are considered the repetition segments. The text used to check for duplications comes from the asset segment, not the fuzzy matches in the TM.
The following token types (if enabled) are treated as interchangeable, and differences are considered equivalent in the repetition analysis, unless auto localization is disabled or the auto substitution penalty is set:
• Dates
• Times
• Measurements
• Numbers
The following segments are included in repetition analysis:
• Untranslated segments with no TM matches
• Segments for which there are only fuzzy matches (whether pre-translated or not)
• Manually translated segments
In order to assure scoring consistency, the following segments are
excluded from repetition analysis:
• ICE matched segments
• 100% matched segments (with or without repair)
• Auto-translated segments
