Custom Machine Translation Management (Custom Glossary Model Details)

What is custom glossary setting/custom glossary model?

The process of wrapping the standard model of MT learning service or the custom trained model registered in this system using a glossary to automatically output the translation defined in the glossary is called “Custom glossary setting” in this system (sometimes called "Glossary wrapping"). As the result of custom glossary setting, a custom glossary model is created. Once the custom glossary setting has been completed, the models set with glossaries can be used in Quick MT/PE of this system.

Information in Custom Glossary Model Details screen

Detailed data of the custom glossary model is displayed on this screen. You can only change the following items.

• Custom Trained Model Name: An arbitrary string of each model. It is displayed in the following places. The default value is “Custom Glossary Model_YYYYMMDDhhmmss" whose end is the timestamp (JST) when the registration of the custom glossary model started. (Example: “Custom Glossary Model_20220201152040" if it starts at Feb 1, 2022, 15:40 JST)
  
  • Quick MT, Quick PE (Project Details screen):
    
    • MT engine "Google Translate": Options in the pull down list of Optional settings [Custom Trained Model]
    • MT engine "Kawamura NMT": Options in the pull down list of Required settings [General/Custom MT Model]
    • MT engine "Amazon Translate": Options in the pull down list of Optional settings [Custom Glossary Model]
• Model Status: Specifies whether to be used in Quick MT/PE. The default value is “Enabled”. Set “Disabled” for a custom trained model that are not used in Quick MT/PE.
• Comment: You can add any information about the custom glossary model.

Also, at the bottom of the screen, the information on the language asset (glossary) used for the custom glossary setting is displayed.

Model evaluation for custom glossary models

On this screen, you can perform model evaluation of the custom glossary model. When you press the [Model Evaluation] button at the bottom of the screen, you can upload a bilingual file (which contains bilingual data of the source text and the correct translation, file format: xliff, csv, tsv, maximum 50,000 characters) for the model evaluation.

As a result of the model evaluation, the following four indicators are calculated as numerical values, and you can see the changes before and after the glossary setting as follows. The first one is the value from the standard model (generic model) (*) of the MT learning service used for the glossary setting, and the last one is the value from the custom glossary model. When a custom trained model was used for the glossary setting, the second value will be displayed which from the custom trained model. If the value of BLUE, NIST, and RIBES changes higher, and the value of WER changes lower, it can be interpreted that an effective customization has been implemented by the custom training. You can also check the detailed meaning of each indicator in this blog post.

• BLEU
• NIST
• RIBES
• WER

Duplicate terms in the glossary

You can register duplicate terms (bilingual data) in the glossary, but it is recommended that a term pair (source-target text) be unique, whether referenced as a glossary or incorporated into a custom glossary model. However, there are cases where you want to change the translation of each term and the translation of a compound word, as shown below.