Content protection: Translation Memory Conversion

Translation Memory Conversion

Challenge: Need to convert TMs

When choosing new import rules for a given language, these will apply for all newly created tasks, but all TM matches with mapped numbers/content will be less than 100% matches. Therefore a solution to convert TMs is needed.

1. When activating an input rule an Information window must pop up: Activation of input mapping rules may lead to the loss of 100% matches in new tasks unless you convert all affected language resources. | Das Aktivieren von Input-Regeln kann in neuen Projekten  zum Verlust von 100%-Matches führen, wenn Sie die betroffenen Sprachressourcen nicht konvertieren.

1. On the content recognition rules and on the input mapping table a new column is needed "Resource conversion | Ressourcen-Konvertierung". For each rule a wrench icon should appear in this column, IF there are still available resources that are not in sync with the active rules. This could happen by defining new input mapping rules (therefore needed on this table) or by disabling/re-enabling a rule in general etc. Add mouseover: Language resources that need conversion. |  Zu konvertierende Sprachressourcen.
   On clicking the wrench icon, the language resource conversion window opens.

1. In the window for the conversion
   list all Memories that need to be converted with a checkbox, that defines, if they are converted, when conversion is started with a click on the "start conversion | Konvertierung starten" button.
   All checkboxes are checked by default.
   The window should contain a grid with the columns "Resource name | Sprachressource", source language | Quellsprache, target language | Zielsprache, assigned clients | Zugewiesene Kunden and should be filterable along those columns.
   There should be an uncheck all possibility.
   Show text: Warning: Conversion cannot be undone without the loss of TM matches. Please backup your resources before conversion. Conversion may affect matches in open tasks, please read how to avoid this on the following confluence page. | Achtung: Die Konvertierung kann nicht ohne den Verlust von TM-Matches rückgängig gemacht werden. Bitte machen Sie vorher ein Backup der Sprachressource. Die Konvertierung beeinflusst die Matches von laufenden Aufgaben. Auf der folgenden Confluence-Seite wird beschrieben, wie Sie dies umgehen können.
2. On the language resource level (in language resource management and in task-language-resource-assoc (wizard and project overview)) we need a new information whether the resource is in sync with all active input mapping rules or not. Warning icon plus mouseover: Content protection rules for this resource have changed, it needs conversion. Please go to the content protection configuration page or ask your project manager (all clients). | Die Regeln für die Erkennung von Inhalten haben sich für diese Sprachressource geändert, sie sollte konvertiert werden. Gehen Sie dazu auf die Konfigurationsseite der Erkennungsregeln oder Fragen Sie Ihren Projektmanager (alle Kunden).

1. On task level, when a resource is used that is not in sync with the active input mapping rules, show the same Warning icon plus mouseover: Content protection rules for this resource have changed, it needs conversion. Please go to the content protection configuration page or ask your project manager (all clients). | Die Regeln für die Erkennung von Inhalten haben sich für diese Sprachressource geändert, sie sollte konvertiert werden. Gehen Sie dazu auf die Konfigurationsseite der Erkennungsregeln oder Fragen Sie Ihren Projektmanager (alle Kunden).

Challenge: How to handle tasks in progress

Ongoing tasks that have been created before input mapping rules have been chosen to become active will lead to match losses.

For now we will not implement a technical solution for this.

The following workaround that will be communicated with beta testers and end users in confluence. Users will be pointed to this confluence page in the pop-up message outlined above.

1. Copy the resource which you would like to convert but is still in use in various tasks, lets call it master_copy. Best and fastest way to copy the resource is to export it in .tm format (not .tmx) and reimport it again under a new name.
2. Convert master_copy
3. Use master_copy for all new tasks
4. Create empty "project TMs" for each ongoing task.
5. Whenever a task is finished, save to project TM
6. Export project TM via tmx and delete TM afterwards
7. Import tmx to master_copy. This applies the content protection rules for the newly imported segments.
8. After last project TM is imported to master_copy rename it master (and delete old, unconverted master)

User can choose not to proceed as described e.g. for unimportant tasks.

For the future

Automatic project-TM creation. For each task a project TM is created with write access. On applying new input mapping rules, a resource can be converted even when it is in use in a task: fuzzy matches will have lower match rates for the ongoing translation but this is only a minor issue and can be ignored. After finishing the workflow, the project TM will be converted according to the active input mapping rules and all its segments will be imported to the already converted master automatically.

This automatic project-TM creation will be available always for all projects (independent of conversion). If it is active, no segments are saved into TMs, that are checked as "writeable", but instead into the project-TM. Project-TMs are not shown in the language resource assoc or management windows, yet their results are shown in the fuzzy match and concordance search windows of a task. A checkbox in the language resource panel (in wizard and at the task) exists, that is called "Use project TM" with a tooltip, that explains what it does. If it is checked by default or not a systemconfig defines, that is  overwriteable on client.