Amazon Translate | Using Amazon Translate Flashcards
What are the most common use cases for Amazon Translate?
Using Amazon Translate
Amazon Translate | Machine Learning
Machine Translation is a great solution in cases where the volume of content is high, speed is critical, and a certain level of translation imperfection (often minor) is acceptable. For example, if you need to extract insights from large volumes of text in many languages, enable customers to search your application in their language of choice, make use-authored content such as forums and support content accessible in languages other than the source, get the gist out of responses to questionnaires and surveys, or publish a first draft – you can use Amazon Translate’s raw output.
With light human post-editing, Amazon Translate can be applied to enabling customer service agents to support anyone, and translating company authored information such as specifications, comparisons of alternatives, FAQs, and support content. With more extensive post-editing, you can also use Amazon Translate to translate high-value, branded content, such as advertising and marketing materials, contracts, etc.
How can I use the service?
Using Amazon Translate
Amazon Translate | Machine Learning
You can call the service directly from the AWS Command Line Interface, or use one of the SDKs in the programming language of your choice to integrate with your applications. Either way, you can start using Amazon Translate for multilingual text capabilities to translate text with just a few lines of code.
You can pass source text to the API and indicate the source and target languages. We return the text translated into the target language. There are three main ways to use the API – first, you can integrate the API into your application to localize highly dynamic application components such as multi-participant chat, for example. Second, you can string it with other services to enable language-independent processing. For example, Database services such as Aurora can be called through Lambda blueprints to enable website localization of moderately-dynamic content such as user generated reviews and forum posts. Finally, you can translate batches of documents. For example, financial services companies can translate and monitor news articles in any language; legal teams can discover materials in multiple languages related to a lawsuit (known as eDiscovery); patent attorneys can search patent repositories anywhere in the world in IP cases.
What kind of inputs does the service support?
Using Amazon Translate
Amazon Translate | Machine Learning
For preview, Amazon Translate takes unstructured text input and language flags to indicate the language of the source text and desired target. If the “language” flag is unknown, customers can use Amazon Comprehend’s language identification API to automatically determine the language, and report that language back along with the translation to the target language.
What are the limits on the API?
Using Amazon Translate
Amazon Translate | Machine Learning
During preview, service calls are limited to 1,000 characters per API call. We provide instructions for how to break up large documents into sections and paragraphs so that customers can translate text of any length.
The Amazon Translate service is highly scalable. During preview, customers may call the service up to once per second per language pair by default. Customers who need higher throughput can request an increase. We will ease both limits significantly when we make the service generally available.
Can I customize my translations?
Using Amazon Translate
Amazon Translate | Machine Learning
Not yet, but we would love to hear from you directly if that is important to you.
Am I required to attribute the translation to Amazon? To Machine Translation?
Using Amazon Translate
Amazon Translate | Machine Learning
You are not required to attribute translations, but we do suggest that you attribute the translation to Machine Translation to inform your own customers.