AWS Comprehend Middleware
The AWS Comprehend Middleware offers Bot Framework middleware components for the AWS Comprehend API. You will need an AWS account, as well as a Comprehend resource created on AWS. Take a look at the Amazon AWS Comprehend documentation for details.
Installing
npm install @botbuildercommunity/middleware-aws-comprehend --save
Authentication
This package assumes that your AWS credentials are set in the environment variables. The following environment variables are required:
AWS_ACCESS_KEY_ID
AWS_SECRET_ACCESS_KEY
Usage
All middleware is created and used in the same way. For example, for sentiment analysis, import the SentimentAnalysis
class from the package, and add it to your bot adapter:
import { SentimentAnalysis } from '@botbuildercommunity/middleware-aws-comprehend';
adapter.use(new SentimentAnalysis());
When used, the turnState
on the TurnContext
will have a property named sentimentScore
between 0 and 1. A full example can be seen in the sample
bot test file.
Differences from Cognitive Services and Watson: AWS Comprehend does not return a 0 - 1 response on the sentiment. Instead, it returns 0 - 1 for each possible sentiment. You can access the sentiment through the sentimentType
property on the turnState
.
By default, the middleware assumes an East US region and English as the language. You can modify this by passing in AWSComprehendOptions
to the SentimentAnalysis
:
import { SentimentAnalysis } from '@botbuildercommunity/middleware-aws-comprehend';
adapter.use(new SentimentAnalysis({ apiVersion: '2017-11-27', region: 'us-east-1', lang: 'en' }));
Supported middleware classes include:
SentimentAnalysis
LanguageDetection
KeyPhrases
EntityExtraction
In each case, properties are added to the turnState
of the TurnContext
. You can retrieve them in your bot via:
context.turnState.get('sentimentScore')
//This is a number for SentimentAnalysis
context.turnState.get('sentimentType')
//This is a string for SentimentAnalysis
context.turnState.get('language')
//This is a string for LanguageDetection
context.turnState.get('keyPhrases')
//This is an array of strings for KeyPhrases
context.turnState.get('textEntities')
//This is an array of strings for EntityExtraction