Correlations Mechanism comprises automated processes which are running in background during translation tasks "Analysis" phase to help decrease translation times as translations of segments which exactly match previous segments in the project can be anticipated and filled automatically. This correlations mechanism even goes further by anticipating future matches from the Translation Memory.


This is different than only using Translation Memory to fill automatically segments. With correlations mechanism, segments are analyzed and linked to a parent segment that has the same source content. They will then reuse translations from their parent segment and be set to "Out Of Scope" status if such a parent segment exists. "Out Of Scope" means that the segment will not be taken into account in the words volume processed by a user, hence not considered in "Purchase Order" data for that user. This mechanism is similar to the concept of repetitions in the translation industry.


You can see how to configure the correlations mechanism scope here: How to configure the Correlations Mechanism Context Scope.


Correlations Mechanism Breakdown


  1. Upon creation, segments of a translation task are analyzed (during the "Analysis" step of the workflow) and compared to existing segments of the project's existing translation tasks first, then to the Translation Memory segments in a second phase. 
  2. Once "Analysis" is done, all the segments which are linked to a parent segment in the project are set to "Out Of Scope".
    Parent segments are the most recent version of the segments among all translation tasks in Archives which contain such segments (if such segment exists), otherwise, they will be the most recent version of the segments from a translation task that is more advanced in the workflow.
    The results of the "Analysis" phase can be fetched for selected translation tasks by clicking on.
  3. Depending on the context scope set in the Settings of a project, these translation tasks can be blocked to prevent them from moving forward in the workflow as they need to wait for the parent segments to be translated first.
  4. When a user that is assigned to a translation task starts processing it, the segments which have a parent segment will be already filled with translations and set to Out Of Scope.
    Such segments can however still be modified and the dependency between the segment and its parent segment will be broken:


Benefits of Correlations Mechanism


  • Deadline anticipation
  • Balanced volume assignation per user
  • Fewer segments to translate
  • Estimate potentially reachable TM matches and anticipations

All of these lead to savings in time and money.


Constraints & Consequences of the Correlations Mechanism


  • Users can work on translation tasks that are locked but cannot move in the workflow translation tasks whose parent tasks are not at least one step ahead in the workflow (e.g a translation task which is in "Review" and whose parent task is also in "Review" cannot go to "Validation" before its parent task).
  • Any update of the content of a parent segment will also update the content of the child segments. This goes only one way: updates in child segments will not update parent segments.
  • The list of translation tasks which a task depends on can be viewed in the task view by hovering the lock icon.


Note that you can still break dependencies between tasks. You can find more information about how to do it here: How to unlock tasks and break their dependencies with other tasks.


It can take time for the system to complete the "Analysis" phase depending on the volume of the source content, the volume of translation tasks in "Archives", the selected context scope, etc.