# Amazon Translate A2I Workflow

Continuously improving AWS translate results using AI to augment parallel data
with humans in the loop. The end goal is to provide customized and accurate
domain specific translation results.

# Solution Architecture

![Solution Architecture](images/solution_architecture.png)

## Prerequisites

1. Download and install the latest version of Python for your OS from [here](https://www.python.org/downloads/). We will be using Python 3.8+.

2. You will need to install version 2 of the [AWS CLI](https://docs.aws.amazon.com/cli/latest/userguide/cli-chap-install.html) as well. If you already have AWS CLI, please upgrade to a minimum version of 2.0.5 follwing the instructions on the link above.

## Deployment Instructions

This solution, as written, will only run within the `us-east-1` region. If you
need to run in a different region you will need to copy the artifacts into a
regional bucket, however the Active Custom Translate feature is only available
in the following regions `us-east-1` (Northern Virginia), `us-west-2` (Oregon)
and `eu-west-1` (Ireland).  Use ![download_artifacts](download_artifacts.sh) to
update the artifacts in your regional bucket.

1. Download the contents of this repository on your local machine (into say:
   `project-directory`)

2. Run ```./download_artifacts.sh <<regional-bucket>> ``` This script will
   download artifacts from blog bucket and upload the artifacts to your
   regional bucket.

3. Using AWS console launch this ![Cloud Formation
   Template](cfTranslateTemplate.yaml). Provide regional bucket name for blog
   bucket parameter.

## Build Instructions

1. Download the contents of this repository on your local machine (into say:
   `project-directory`)

2. The solution is implemented in python, so make sure you have a working
   python environment on your local machine.

3. Build the package using ./package.sh script. This will generate
   translation-helix.zip file.

4. Optionally build the nlt-layer using "build-nltk-layer.sh" in an Amazon
   Linux2 OS. For building lambda layer the libraries getting packaged must be
   compatible to Amazon Linux2. Use Cloud9 or EC2 instance with Amazon Linux2
   to complete this task or you can download the nlt-layer.zip from this
   ![location](https://aws-ml-blog.s3.amazonaws.com/artifacts/amazon-translate-a2i-workflow/nltk-layer.zip).

5. Use the CloudFormation template. Navigate to the src sub directory. Package the contents and prepare deployment package using the following command
    1. ```bash
        aws cloudformation package --template-file cfTranslateTemplate.yaml --output-template-file translate-a2i-setup-output.yaml --s3-bucket <<your-bucket>> --region us-east-1
        ```

## Execution Instructions

To execute the workflow, follow the instructions in blog post.


## Further Reading:

- Blogpost: [Designing Human Review workflows using Amazon Translate and Amazon Augmented AI ](https://aws.amazon.com/blogs/machine-learning/designing-human-review-workflows-with-amazon-translate-and-amazon-augmented-ai/)

## License

This library is licensed under the [MIT-0 License](https://github.com/aws/mit-0).