Translation Memory and Glossary

The Translation Memory and Glossary tab lets you manage your linguistic resources. Here you can decide which resources should be used for a project (both before and after creation), adjust how they’re used (to look up matches, store them or both), and adjust settings such as penalties and priority. You can also control how strictly Matecat matches language variants to make sure the suggestions fit your project’s needs.

The Translation Memory and Glossary tab has two separate lists:

Active Resources lists the resources that you have assigned to the project you are creating or working on and MyMemory, the public translation memory shared by all Matecat users, which is always shown at the top of the list of active resources and is marked by a globe icon. When no private resources have the “Update” box checked, all translations are automatically saved inside the public memory.

Private resources can be activated by clicking the “Activate” checkbox next to the resource’s name. Upon activation, private resources will have both the “Lookup” and “Update” boxes checked, which means that they will be used to both look up existing translation matches and store new ones. Whenever at least one private resource has the “Update” box checked, new matches won’t be stored in the public memory. For each private resource, you can decide to deactivate one of the functionalities by unchecking either of the “Lookup” and “Update” boxes.

Only Lookup checked: the resource is only used to look up existing translation matches and glossary terms.

Only Update checked: the resource is only used to store new translation matches and it is possible to add new glossary terms to it;

Both Lookup and Update checked: the resource is used both to look up existing translation matches and glossary terms, and to store new matches.

Inactive Resources lists all the resources linked to your account that have not been activated. As mentioned above, you can activate a resource by clicking the “Activate” checkbox next to its name.

Penalizing resources

Every resource, including the public one, can be penalized.

To apply a penalty to a resource, click its corresponding “Add penalty” button and set the penalty value. The match percentage from the penalized resource will be reduced according to the assigned penalty value.

If a 5% penalty is applied to a resource, a 100% match from that resource will be treated as a 95% match in both the analysis and the editor.

This means that, in the analysis, the match will be placed in the “TM Partial 95-99%” bucket, rather than in the “TM 100%” one. Additionally, in the editor page it will appear after any matches with a match percentage greater than 95%.

In the editor, the penalty percentage is shown next to the TM match. By hovering the mouse over the penalty percentage, a tooltip with more information will pop up.

Prioritizing resources

By default, if two matches from different resources have the same match percentage, the most recent one appears higher in the list. By activating the "TM prioritization" option, this behavior changes: in cases of matches with the same percentage, those from higher-priority resources are shown first. Priority is determined by the position of a resource in the list: the higher the position, the higher the priority.

In the case of multiple 100% or 101% matches from different resources, priority determines which match is applied during pre-translation. If two 101% matches are present in the resources, the one from the resource placed higher in the active resources list will be used for pre-translation.

Variant-strict matching

To maximize TM leverage, Matecat, by default, returns TM matches from all variants of the job’s source and target languages. Matches from language variants other than those specified for the job incur a 1% penalty to prevent their use in pre-translation. For example, if your project is English (UK) to Spanish (Spain), Matecat will also show matches for the following combinations:

English (UK) to Spanish (Mexico)

English (US) to Spanish (Spain)

English (US) to Spanish (Mexico)

When variant-strict matching is enabled, Matecat only provides variant-specific matches. In the example above, it would only return matches for the English (UK) to Spanish (Spain) combination. To activate variant-strict matching, check the relevant box in the tab.