Features available with the Language IO API

The purpose of this article is to describe the features that are accessible through the Language IO API, and how to best leverage these features against your specific needs. It also shows the tradeoffs that exist when choosing to use extra features in requests (latency, monetary cost, etc.).

Features available with the Language IO API

Smart Model Selection (SMS)

Enabled by default: Yes.

Smart Model Selection is a process that takes translation quality feedback data from all Language IO customers to continuously evaluate the top-performing machine translation models by language pair. This means that the feedback for each language pair is evaluated independently. This data is then used to adjust both global and per-customer engine priority.

Smart Model Selection is enabled by default for all new accounts. This includes API demo accounts. For more information, see Understanding Smart Model Selection.

Why should I use it?

• Smart Model Selection guarantees that you always use the best available engine at all time, automatically.

What's best for my use case?

• You may request specific engine preferences to be configured. However, this means that you would no longer be eligible for SMS evaluation. In other words, while this gives you greater control over the engines you want to use, disabling SMS means that you lose the flexibility and optimization of your engine options. This may impact response times or translation quality.

Glossary Imposition

Enabled by default: Yes.

The Glossary Imposition feature searches for certain terms in a text to translate or replace them with preferred terms in an existing glossary. This can be for example to make sure a term is consistently translated or to ensure, as applicable, that preferred terms are always used, or conversely that undesired terms are never used.

For more information about how to build a glossary for Machine Translation, see Creating a Glossary for Machine Translation: Best Practices 

For more information about Glossary imposition, see Glossary imposition in translations.

Why should I use it?

• Glossaries improve the consistency and quality of the delivered content as they make sure your preferred words or expressions are used in all translations.

What's best for my use case?

• Because of the added processing of searching and replacing, Glossary Imposition may cause increased latency. This means that, if your priority is speed, using glossaries may not be the best option for you. 

Self-Improving Glossary (SIGLO)

Enabled by default: No, opt-in feature.

The Language IO Self-Improving Glossary (SIGLO) is an automated process that runs in parallel to your translation requests, without interfering with or changing your workflow, to flag new or improved terms that might warrant an inclusion in your glossary.

For more information, see Understanding the Language IO Self-Improving Glossary (SIGLO).

Why should I use it?

• The SIGLO process helps improve your glossaries dynamically over time, using real-world data and feedback.
• Because the process runs in parallel to regular translation requests, it does not affect the latency of deliveries. 

Enabling SIGLO

• SIGLO is an opt-in service for Language IO customers. If you think you could benefit from SIGLO, submit a support request.
• SIGLO is not available for API demo accounts.

Translation Optimization and Optimization Quality Estimate

Enabled by default: Yes, English only.

When Optimization Quality Estimate is enabled, outgoing messages pass through a Large Language Model routine that lightly alters the source content to increase the likelihood of a high-quality machine translation. For example, this can make a long message more direct and, to a certain extent, through this rephrasing, fix typos that were in the original text.

For more information, see What is Optimization and how does it work?.

Why should I use it?

• Optimization helps improve the quality of translations because it cleans up and makes the source content simpler.

What's best for my use case?

• Optimization is a parameter in the Translate API Route. Its use is optional.
• Optimization is only available for English content at this time. If your primary source language is not English, you cannot use this parameter. 
• Because Optimization needs to send additional requests to work, its use incurs additional costs in credit consumption and latency. This means that, if your priority is speed, using Optimization may not be the best option for you. 

Translation Quality Estimation (TQE) and Quality Model Failover

Enabled by default: Yes.

Translation Quality is a process that evaluates the quality of machine translations. If a translation falls under the defined quality threshold, the system uses a second machine translation engine to do a retranslation automatically (This is the Translation Quality Failover). The second translation is scored and the score is compared to that of the original translation. The translation that has the higher score is delivered.

For more information, see Understanding Engine Failovers.

Why should I use it?

• The TQE process makes sure that the best possible translation is always delivered, based on your own quality criteria. 

What's best for my use case?

• When enabled, TQE is a parameter in the Translate API Route. Its use is optional.
• Using TQE can incur additional costs in credit consumption (for automatic retranslations) and latency. 

Attachment Translation

Enabled by default: On some accounts. If your Language IO subscription started before November 19th, 2024, this new feature requires your organization to sign a new contract. Contracts created after November 19th, 2024 have the feature enabled by default.

See Using the Language IO API to translate files.

Why should I use it?

• Use Attachment translation if your end users often send supporting documents such as PDFs or images.

What's best for my use case?

• Attachment translation is an optional feature of the Translate API Route. 
• Translating files can take up to a few minutes, depending on the size of the file (compared to a few seconds for plain text). This means that, if your priority is speed, implementing the option to translate attachments may not be the best option for you.