Translation memory (TM): Frequently Asked Questions (FAQ)

Introduction

The purpose of this article is to answer common questions about the Translation Memory Management (TM) module in XTM Cloud and to help resolve issues you might experience with it.


Questions and answers

What do prevCRC and nextCRC mean in a sample import file?

The nextCRC and prevCRC fields contain information about a segment’s context. They point to specific trans-units. These create the context for the TM module, enabling this context to be imported or exported. This is used to obtain ICE matches from the imported segments.

I translated my content from English (USA) → German (Germany). Is there a way for me to now use it for a German (Germany) → English (USA) project?

Yes, this is possible. You can use what is known as reverse TM matching. As a user with the Administrator role, select ConfigurationSettingsTranslationTM Matches - general. Then select the Reverse memory matching option. This enables the swapping of the source and target language in a particular translation memory.

Untitled29.png

Can I insert a translation unit (TU) ID in the export/import file?

Currently, it is not possible to add parameters of this kind to a TM import file, in XTM Cloud.

Import of the TMX import file failed with errors. How can I determine which segments were not imported and why the import failed?

Such errors are nearly always due to inline differences. To arrange to have the file verified, and find out why import failed, raise a ticket with the XTM International Support team, on our Support Portal.

What does the Use user's language combinations option mean?

The Use user's language combinations option is located in a user’s Translation Memory access rights [UsersUser list → (select a user) → Access rightsTranslation Memory). If this option is selected, the user cannot check every language in the TM. The user can only search for assigned language combinations. If this option is not selected, the user will see all configured languages in Concordance and TM. To configure these languages, select ConfigurationDataLanguage combinationsCustomize.

Untitled30.png

When I change the Matches - general settings, do my changes apply to projects that have already been created or are in progress or only for the projects that are created/reanalyzed after the changes have been applied?

To make the Matches - general settings, select ConfigurationSettingsTranslationTM. Changes related to matching are only applied to a project when it goes through matching, either during creation or reanalysis. These changes will not be applied in projects that have already been created unless they are reanalyzed.

Untitled31.png

Should ICE matches from not approved TM be counted in the project’s Metrics, under ICE match?

To select the Use not approved memory option, select Project EditorGeneral infoGeneral info.. If you select this option during project creation, not approved TM ICE Matches will be shown in the project’s Metrics, under ICE Match.

Untitled32.png

Can I search for a TM entry with an apostrophe?

Currently, the XTM Cloud TM search does not support searching for characters other than letters and numbers. This restriction is necessary to avoid performance issues if a single character such as . or , is searched for. A search like this would result in a lot of TM entries being found. If it was performed by multiple users at the same time, this would negatively affect the performance of the server.

However, note that the search can be also performed with a Phrase match instead of an Exact match. If this was combined with the option to perform a Case sensitive search, the results would be similar to those based on the source text:

Untitled33.png

What is the default order sequence for TM matching in a project? Can I change it somehow?

Whenever segment ID is being used, it takes priority over the ‘'normal matching rules’'. Segment ID matching can be configured in a particular filter template, per customer or at project template level. ICE matches can be penalized so that they only leverage matches at project level. Other than that, depending on whether or not a particular TM match exists, the default order of matching in a project is as follows:

  1. Segment ID.

  2. In-context exact (ICE) matches.

  3. Repetitions.

  4. Leveraged matches.

  5. Cross-file repetitions.

  6. Fuzzy matches.

  7. Machine translation (MT) matches.

  8. Fuzzy repetitions.

Keep in mind that this order can always be changed using the TM settings (global/customer-level/project-level) or TM penalty profiles, for example.

When I was working on a project, Fuzzy matches etc., all became ICE matches or Repetitions after it was moved to the Correct step, from the Translate step. Is this expected behavior?

Yes, this is expected behavior. Matching can change during the project, or after it has been reanalyzed. As XTM Cloud actively checks the TM database all the time, new TM matches can emerge.

Is it expected that the metadata for a match does not show the name of the Linguist when an LSP is assigned?

Yes, this is expected behavior. The first number is the LSP ID and the second is the user ID on the LSP side. You will only see those IDs when anything is assigned to and translated by the LSP.

How can I delete a TM’s contents?

If you select the relevant language combination in the TM → Manage section, perform a search, and click on Delete all, you will delete everything found in the search, and all corresponding data will be deleted. At XTM International, we use the same procedure.

Untitled34.png

Can I apply penalties on reverse matching?

By default, there are no penalties on reverse matching. However, depending on what you would like to achieve, this can be implemented by configuring TM penalty profiles in the appropriate way.

Can XTM Cloud automatically check differences between source texts and target texts in an existing TM, or something like that?

There is no possibility to check the records in TM in bulk. However, you can perform the check for individual records, using the Check current segment option.

Untitled35.png

I wanted to apply machine translation (MT) in a project but I see Fuzzy matches from TM instead. How can I arrange for only matches from MT to be applied to the project?

If population with TM matches (the application of translation matches from the TM, in target segments) is enabled, all the standard TM matches always have priority over MT matches. This is expected behavior, implemented based on the results of statistics proving that Fuzzy matches are of higher quality than MT matches.

If you prefer to use MT matches instead of TM matches, disable Fuzzy match population. To do so, select ConfigurationSettingsTranslationTMFuzzy matches (administrative privileges required).

Untitled36.png

IMPORTANT!

Keep in mind that the settings above are global settings and will affect all projects in XTM Cloud, not only those sent for machine translation. Additionally, if you wanted to have no TM matches applied at all, the project would need to be created under a new customer (that does not have TM).

Repetitions are not populating correctly. Is that correct? Is the default not that, when the translator translates a repetition’s parent, the system will automatically propagate it to all child segments.

You are correct: the system does work like that. However, a TM match can only be displayed and matched to a particular segment, in the project from which it originated, when repetitions are translated from the bottom to the top. In such a case, XTM Cloud will check the TM and the first occurrence of the repetition will be matched with the TM.

To learn about cross-file repetitions in relation to TM matching, see Cross-file repetitions and TM matching.

Why has a Fuzzy match been accepted (set to “Done”) in a segment in XTM Workbench?

In general, XTM Cloud only accepts Fuzzy matches in one case – when the Change segment status automatically option has been enabled in XTM Workbench. It is enabled personally, for a specific user. To enable it, select SettingsGeneralSegments.

When the Change segment status automatically option has been enabled, and you leave a target segment, its status automatically changes to “Done”. “Done” segments are also saved automatically in the translation memory. Furthermore, if the workflow step in which the segment is set to “Done” has the Approve TM option enabled, the TM is updated and marked as “Approved”.

Untitled37.png

Why is half of the project locked and there are a lot of unapproved TM matches?

The most likely reason is that the Allow to edit the ICE segments option is not selected for the project and the Use not approved memory option is selected, which permits the use of not approved TM matches in the project (Project EditorGeneral infoGeneral info).

Untitled38.png

Why does the segment status change when a Linguist clicks on the segment? Some segments are displayed as ICE matches on the bottom left screen, the Matches section or in the Concordance section, with the current project number and date.

Those TM matches come from a different file in the same project (this project).

Keep in mind that TM matching for the project is not only performed when the project is analyzed but also per segment, every time that the segment is activated in Edit mode in XTM Workbench, which is also why the segment status changes if the segment is activated. Therefore, internal TM matches must have affected the project matching in general.

How do I batch approve the TM from the project?

One way is to select the Approve TM option in a particular workflow step in which the translations are performed:

Untitled39.png

Alternatively, you can add Approve TM as a separate step (one of the default automatic steps), so that the project’s translation memory will be approved after the project has been fully processed.

Untitled40.png