feedback_sentiment_analyzer

Feedback Sentiment Analyzer (FSA)

Feedback Sentiment Analyzer (FSA) is an example application that showcases AWS services and SDKs. Built with ❤️ for you to explore, download, and deploy.

What it does

Specifically, this application solves a fictitious use case of a hotel in New York City that receives guest feedback on comment cards in a variety of languages.

These comment cards are uploaded through a web client, transformed using a suite of machine learning services, and rendered through the same web client. The end result contains the original image, an English translation of the text, and an audio element.

---

SDK implementations

This application has been implemented with the following AWS SDKs.

• Ruby
• JavaScript
• Java
• .NET

To deploy one of these implementations, follow the Deployment instructions.

---

AWS services used

This application uses a suite of AWS machine learning services to do the following:

• Extract text using Amazon Textract
• Detect sentiment using Amazon Comprehend
• Translate to English using Amazon Translate
• Synthesize to human-like speech using Amazon Polly

Additionally, the application showcases the following AWS services:

• Amazon Simple Storage Service (Amazon S3) - Stores images of comment cards
• Amazon EventBridge - Relays events from Amazon S3
• AWS Lambda - Executes business logic
• AWS Step Functions - Orchestrates multiple Lambda functions
• Amazon DynamoDB - Stores details about each comment
• Amazon API Gateway - Routes requests from frontend to backend
• Amazon CloudFront - Distributes this application globally
• Amazon Cognito - Authenticates users

---

Deployment instructions

This application is deployed using the AWS Cloud Development Kit (AWS CDK).

1. Get AWS credentials.

2. Set the following environment variables:

   • FSA_NAME - Any text less than 10 characters
   • FSA_EMAIL - A valid email address that you own
   • FSA_LANG - Any of the implemented languages

   For example:

   Bash

   export FSA_NAME=ana
   export FSA_EMAIL=ana@example.com
   export FSA_LANG=ruby
   

   Windows cmd

   set FSA_NAME=ana
   set FSA_EMAIL=ana@example.com
   set FSA_LANG=ruby
   

   Windows Powershell

   $Env:FSA_NAME = ana
   $Env:FSA_EMAIL = ana@example.com
   $Env:FSA_LANG = ruby
   

3. Run the following commands:

   cd cdk
   npm install
   cdk deploy
   

   After deploying, observe the Output in your terminal session. Copy the CloudFront distribution URL, which has websiteurl in the name. Paste this URL into a browser to launch the application.

---

Application instructions

Sign in

1. Choose Sign in.
2. Enter the the email from FSA_EMAIL into the Username field.
3. Enter the temporary password that was sent to you into the Password field.

4. Reset your password as prompted.

Upload images

1. Signing in directs you to a landing page that says No data found.

2. Choose Upload.
3. Choose Select a file.

Note: Feel free to choose a sample comment instead of writing your own. 4. Select a PNG or JPEG image that contains a positive comment about the hotel. Negative comments are saved, but not returned to the frontend.

5. Choose Upload.

Play the result

1. Wait. The image upload takes a minute or two to process. CloudFront will also cache API responses for five minutes. To speed things up, create a CloudFront invalidation.
2. Choose the Refresh button.

3. You should see a card with the original image, translated text, and a play button.

4. Choose the play button to listen to the translated audio.