Translation Memory (TM) Optimization allows you to create an optimized copy of an existing TM or merge multiple TMs into one optimized copy. Traditionally, TM cleanup required a human linguist, but with Smartling’s TM Optimization feature, this process can now be fully automated. During TM optimization, noisy, unreliable, outdated or inconsistent data is identified for removal or correction based on fully customizable criteria.
TM Optimization is currently available as a beta feature. If you're interested in optimizing your TM for translating with Google Adaptive, please contact your Smartling Customer Success Manager to have this feature enabled in your account.
When should TM Optimization be performed?
To use Google Adaptive as your Machine Translation provider, a dataset of existing translation examples is required to train the engine on your preferred tone, style, and voice.
Smartling’s Translation Memory (TM) Optimization can be used to produce a clean dataset for translation with Google Adaptive in TMX format. Therefore, it is highly recommended to perform TM Optimization before starting to translate with Google Adaptive.
Info: Please note that this beta version of Smartling’s TM Optimization feature is intended primarily for use with Google Adaptive. We do not recommend using an optimized TMX file for direct leverage in your Smartling projects, as it does not contain any of the original metadata, such as the original project, variant, translation date, or translator name. TM units in the optimized TMX file are entirely new units and are not tied to the original project strings in Smartling.
How It Works
Smartling’s TM Optimization functionality allows you to create an optimized copy of an existing Translation Memory. The existing Translation Memory will remain unaltered. It is also possible to select multiple TMs to combine into a single optimized TM.
Key Features
- Customization: Choose specific optimization criteria to determine which entries should be removed, repaired, or flagged.
- Multi-TM Support: Combine multiple TMs into one optimized dataset.
- Google Adaptive Compatibility: Use the optimized TM as a clean dataset for translation with Google Adaptive.
Translation Memory Units
TM Optimization is centered on analyzing the contents of your TM. Your TM is made up of translation units (TU, also referred to as a unit) which consist of source text and its translation in a specific target locale. You can specify optimization criteria when submitting a new TM optimization request. These criteria are then used to remove, repair, or flag specific units.
You can view a translation unit's history and metadata by clicking on the Activity icon. For more information see Translation Units in the Translation Memory.
Important Considerations
- Original TM Preservation: The original TM remains unchanged and can still be referenced to access specific metadata.
- Metadata Limitations: The optimized TM will lack unit metadata, such as the original project, translation date, and translator name. This is because the process generates entirely new translation units.
- Editing Restrictions: Features like “Save & Update Projects” are unavailable, as the new units are not linked to the original project strings.
- Use with a Linguistic Package: The optimized TM will not automatically replace existing TMs in your Linguistic Package. It is intended to serve as a clean dataset specifically for translating with Google Adaptive.
How to Create an Optimized TM
- From your Account Settings, navigate to Linguistic Assets.
- In the menu on the left side, select Translation Memory.
- Next to the Translation Memory you want to optimize, click Optimize TM.
- If you would like to create one optimized TM from multiple TMs, select the desired TMs, then click Actions > Optimize TM.
If selecting multiple TMs, they must share the same source language. It is not possible to optimize multiple TMs with different source languages.
- This will open the five-step optimization request page. Follow the steps below to set your optimization preferences. You can exit at any time by clicking Cancel at the bottom left.
Language Selection
You will see a summary of the total number of translation units and words. Under Target Languages, choose which languages in the Translation Memory you want to keep and optimize.
For more detailed information on the preferences discussed below, please see TM Optimization Options.
Cleaning Preferences
Under Cleaning Preferences, specify which translation units or metadata from units should be removed during the creation of the optimized TM copy.
- From the drop-down menu, you can either use Smartling's pre-configured recommended settings, or create a custom configuration.
- Tick or untick any of the available criteria based on your requirements.
- Click Next to confirm.
Repair Preferences
Under Repair Preferences, specify which repair options will apply to your optimized Translation Memory. This allows you to benefit from Smartling's advanced capabilities to self-heal corrupted translation units.
- From the drop-down menu, you can again choose between Smartling's pre-configured recommended settings or a custom configuration.
- Tick or untick any of the available criteria based on your requirements.
- Click Next to confirm.
Flag Preferences
Under Flag Preferences, specify which mismatched units should be flagged for review. After the optimization process is complete, you can download a summary report to review the flagged units.
- From the drop-down menu, you can again choose between Smartling's pre-configured recommended settings or a custom configuration.
- Tick or untick any of the available criteria based on your requirements.
- Click Next to confirm.
Review & Submit
Lastly, you will be asked to review and confirm your selections. This screen allows you to view a summary of your TM Optimization request.
- Once you have reviewed your optimization configuration, click Submit to initiate the TM Optimization process.
The process of creating an optimized TM may take an hour or more. You can safely navigate away from this page, you do not need to keep this page open during processing.
View Optimization Requests and Summary Reports
You can view all your completed TM Optimization requests by:
- From your Account Settings, navigate to Linguistic Assets.
- In the menu on the left side, select Translation Memory.
- Select the tab Optimization Requests.
Click the View Summary button to access the Optimization Summary. Here, you can see the number of units that were deleted, modified, or flagged. If any units were flagged, you can download a CSV report listing the flagged units.
Export Optimized TMX Files
You can download an optimized TMX file to use as a dataset for translation with Google Adaptive.
- From your Account Settings, navigate to Linguistic Assets.
- In the menu on the left side, select Translation Memory.
- Select the tab Optimization Requests.
- Click the ellipsis menu on the right, and then Export as TMX file.