Information in MT Leaning Service Settings screen

On this screen, you can apply for the use and suspension of MT learning services necessary to use LAC. Currently, the following four MT learning services are available, and the models that can be created are different for each service. The LAC Administrator(s) of your organization must check the terms of use for each service (or the notes on the MT learning service below), and then apply for use ([Apply to use] button).

When you press the [Apply to use] button, the terms of use of each service (or a link to a page with the relevant information) will be displayed. Please read the contents carefully, and if you agree, check the [I agree to the terms of service] check box and click the [Apply to use] button. You can use the services you have applied for immediately.

Notes on MT learning services

An overview of the three MT learning services currently supported by LAC is as follows.

• Model creation/operation environment: The environment in which the model customization process (custom training or custom glossary setting) is actually performed, and the location of the model's data. For “Kawamura NMT”, the model customization process will be done by NICT, and the subsequent operation will be carried out by Kawamura International Co., Ltd.
• Provision of your data: Only “Kawamura NMT” provides your language asset data inputted as corpus for the custom training to Translation Bank. Translation Bank is the name of a project launched by the Ministry of Internal Affairs and Communications and NICT (National Institute of Information and Communications Technology) in September 2017. The provided data will only be used by NICT to train the MT service, and the source/target text will not be reused externally as it is. For more information about Translation Bank, please click here (Japanese only).

• Custom trained model lifecycles: When the MT learning service used to create the custom trained model is released with a new version, the custom trained model created in the previous version will be invalid for MT after a certain period of time has elapsed. The custom glossary  model created by setting a glossary to such custom trained model will also be invalid as the same. Please recreate the custom trained model by performing a custom training again before (or after) the end of the lifecycle.

Stop using MT learning services

You can stop using the MT learning service you are using by clicking the [Apply to stop] button in the list. Please note that if a custom MT model (custom trained model, custom glossary model) created using the service already exists, the following effects will occur.

• LAC (Custom MT Model Management Menu): It will no longer appear in the model list.
• Quick MT: If the data of a model is deleted, it will no longer appear in the pull-down list of the available models for the relevant MT engine.
• Quick PE: If the data of a model is deleted, it will no longer appear in the pull-down list of the available models for the relevant MT engine. Also, if you have a project configured with the deleted model, you can reopen the File Editor screen for that project, but re-translation will result in an error.