Configuration.md

Azure Health Bot: Consuming Cognitive Services Demo

Introduction

We need to perform following tasks to successfully call any API from Health Bot.

• An API you would like to consume. You can consume someone else's API or an API developed for you as well. We will consume Azure Cognitive Services TextAnalytics or preview link and Question Answering APIs.

• Azure Health Bot from which we'll consume APIs. We will create data connection to consume following APIs:

  • Entity Recognition,
  • Entity Linking,
  • Recognize PII,
  • Key Phrase Extraction,
  • Sentiment Analysis,
  • Language Detection,
  • Answer Questions (Static) - using question embedded in the API call,
  • Answer Questions (Dynamic) - by passing user supplied question to the API call.

  Note

  • The first six services are part of Azure Language Service
  • The last two are part of Azure Question Answering service.

Create Language Studio instance in Azure

• Quickstart: Get started with Language Studio

Create Question Answering instance in Azure

• Quickstart: question answering
• For step #10, use the following urls instead of Surface Book User Guide.
  • Azure Cognitive Search FAQ: https://docs.microsoft.com/azure/search/search-faq-frequently-asked-questions
  • [Optional] FAQ for Azure Cognitive Services containers: https://docs.microsoft.com/azure/cognitive-services/containers/container-faq
  • [Optional] FAQ for Azure Cognitive Services disconnected containers: https://docs.microsoft.com/azure/cognitive-services/containers/disconnected-container-faq

Create Azure Health Bot in Azure

Go through these steps to get familiar with the Azure Health Bot:

• Create your first Health Bot
• Experience a built-in Health Bot conversation
• Author your first Health Bot scenario Quickstart

Now, you should be familiar with the Health Bot. You should be able to create custom scenarios by adding prompt, statements etc. This is nothing but a workflow based IDE (Integrated Development Environment) where you can execute code behind the scenes without writing any!

Create Language Service (Text Analytics) instance

• Login to Azure Portal

• Create a New resource.
  

• Select Language Service resource.
  

• Select Language Service features as desired. For demo, select all.
  

• Supply required subscription, resource group, region, pricing details and name for your resource. For demo, Free tier is sufficient.
  

• You can supply additional details in Network, Identity and Tags screen if needed.

• Before you click on Review + create, you need to accept Microsoft's Responsible AI terms and conditions. Click Microsoft Responsible AI principles in practice for more details. You can also review service agreements on the Review + create page.
  

Language Service TextAnalytics Configurations

Now, let's save a couple of the key settings we'll need for our demo. They are

• Subscription Key: We will use this option to authenticate API calls. For production choose sophisticated authentications like AAD or equivalent authentication mechanism.

• Endpoint url: The service url for making API call.

Note: Save subscription Key and Endpoint Url in notepad for later use.

Next

Import sample to Azure Health Bot