Setting Up Translation Memory Fill-up

Turn on Translation Memory fill-up and automatically use 100% TM matches as translations.

Jacob avatar
Written by Jacob
Updated over a week ago

📝Note:

  • Translation Memory Fillups feature is available on the Growth plan and up.

  • Fuzzy Translation Memory Fill-up (Translation Memory Fillups for less than 100% match) feature is available on the Enterprise Plus plan and up.

You can automatically use 100% Translation Memory matches as translations whenever new content is uploaded. This will save you from needing to retranslate those strings or copy TM suggestions into the translation area in the Editor, especially if the strings are pluralized.

📝Note: For pluralized strings, the fill-up feature operates in the following way:

  • Every plural form should have a minimum of a 100% match.

  • Translation Memory suggestions should be linked to previously translated pluralized strings for target languages with just a single plural form, such as Chinese. Suggestions associated with non-pluralized strings will be returned in the editor as suggestions but won't be utilized during fill-ups.

To enable Translation Memory fill-up:

  1. From the Dashboard, click on the project you want to enable TM fill-up for

  2. Click on Settings

  3. In the submenu, click on Workflow.

  4. In the Pre-translation step, enable the Translation Memory Fill-up option.

    workflow-tm-auto-fill-up.png#asset:4281

  5. Click the Save Changes button at the top of the page

The autofill process may take several minutes to run after uploading a new file or updating an existing one.


Fuzzy Translation Memory Fill-up

You can automatically use Translation Memory entries as translations even if there is less than a 100% match; this feature is called Fuzzy Translation Memory Fill-up. It allows you to use your past translations even more and auto-translate your strings with Translation memory entries that are similar but not identical (fuzzy matching).

📝Note: Please note that fill-ups with less than a 100% match will not be applied for pluralized strings.

To enable TM fill-up with less than 100% match, please check the respective option under your project settings in the Workflow tab in the Pre-translation step. You will need to specify your preferred threshold. The auto fill-up will ignore TM entries with leverage below that threshold.

As long as the setting "Translation Memory Fillup for less than 100% match" is enabled, auto Translation Memory fill-up will try to find the best match following the priorities below:

  • similarity %

  • context match

  • language

  • last used

Any translation pair that came from an auto fill-up with less than 100% will not enter Translation Memory until it gets reviewed.

Information about the similarity/leverage percentage is shown in the source string area in the editor.

Screenshot-2022-03-21-at-7.13.37-PM.png#asset:10095


Auto-Review and Proofread after TM Fill up

📝Note: Auto-Review and Proofread after TM Fill-up feature is available on the Enterprise Plus plan and up.

You can select whether to automatically review or even proofread translations from Translation Memory entries that matched 100% or 101%.

We offer the option to select the final status of the translation string to be either reviewed or proofread. Additionally, you can define the threshold for the auto-review by the option "Automatically review translations when TM match is at least":

  • 100% (includes both 100% and 101% matches)

  • 101% (includes only 101% matches)

The above threshold can also be configured for auto-proofread, provided you have enabled the proofread step in the project workflow settings.

Setting up the auto-review or auto-proofread option will save you time and resources since you will not need to order reviews from your LSP provider or engage internal reviewers.

A successful auto-review following TM Fillup will generate the detailed logs below in the string history tab:

Respectively, a successful auto-proofread following TM Fillup will generate the detailed logs below in the string history tab:

📝 Note: When the auto-review or auto-proofread option is enabled, the transifex-bot will be displayed as the reviewer in both the string history and reports.


Triggering Autofill Tasks

Autofill Tasks are the two pre-translation tasks that you can set up in the Workflow tab for a project:

  1. Translation Memory (TM) Fill-up

  2. AI & MT Fill-up (available on Growth Plan and up)

Autofill tasks will run whenever one of the following events occurs:

  • A new translation is added (by any method - through API, CLI, or manual file upload)

  • A new TMX file is uploaded.

  • A source file is updated.

  • A new project is added to a TM group.

  • Custom TM Groups are enabled/disabled.

  • already excluded languages in TM Fill-ups are removed from the list

  • TM/AI & MT fill-up setting is enabled.

  • An MT translation is reviewed.

  • The limit for Translation Memory Fillup for less than 100% match is increased.

  • A translation that came from a less than 100% auto TM fill-up gets reviewed.

  • A target language is added to a Project.

  • A translation is reviewed while "Allow only Reviewed translations into the Translation Memory" is enabled.

  • new strings are approved for translation through the TX Live sidebar

📝 Note: TM and AI & MT auto fill-ups won't run automatically for strings in TX Live that were approved before either of the settings was enabled.


Being notified when TM fill-up is complete

The time required for the Autofill tasks to be completed depends on the data that needs to be processed by your project’s Translation Memory. Please remember that on the Growth plan and up, new phrases and TMX files uploaded to Transifex are processed by the TM first to leverage the previous work quickly.

We highly recommend using Transifex Webhooks to get notified when the TM fill-up tasks are completed.

If Webhooks is not an option for your team, then other options are:

  • Create automated tasks that pull the translations, for example, 15-20 minutes after the source files are updated.

  • Get resources' statistics via API as described here and decide, based on a certain translation completion (100%, 80%, etc.), whether the translation files should be pulled.


Identifying strings translated by TM

There are two ways to know how many words were automatically filled up by TM.

Filter in the Editor

You can identify the strings auto-filled by TM in your resource files by applying the following filter: More > Origin > TM.

Then, you can mass-select all filtered strings and see the word count.

You may also specify the percentage leverage. This is useful if you allow Translation Memory fill-ups with less than 100% match.

tm_leverage_filter.png#asset:9934

Reports

You can see the number of words automatically filled in by TM on the Reports page.


💡Tip

Looking for more help? Get support from our Transifex Community Forum!

Find answers or post to get help from Transifex Support and our Community.

Did this answer your question?