QnA Maker is a cloud-based Natural Language Processing (NLP) service that enables developers to build, train and publish a sophisticated bot using FAQ pages, support websites, product manuals via a simple user interface or REST APIs. The ability to create and publish a bot without writing a single line of code helps in saving training costs for the organisation.
In addition to no code, Automatic extraction,Multi-turn conversations and scaling as per need provides an an opportunity to organisations to create a standalone system to cater to different needs of the company.
In this post, we will create a QnA Maker Service, a knowledge base and QnA Bot to interact with each and provide answers to FAQs entered by the user.
- An Azure Subscription
Create QnA Maker Service
- Login to Azure Portal
- Search for “QnA Maker” and Click on Create
- Provide details and Click on Review + Create
Create QnA Maker Knowledge Base
- Sign in to the QnAMaker.ai portal.
- Select Create a knowledge base.
- On Create a Knowledge Base Page , Skip Step 1 as we already created QnA Maker service in previous section.
- Step 2: Select the QnA Maker service and Language in the drop-downs
- Step 3: Provide a Name for your knowledge Base
- Skip Step 4 as we would add QnA Pair after creating the knowledge bas.
- Step 5: Click on Create you KB
Add QnA Pair to QnA Maker Knowledge Base
- Click on Add QnA Pair
- Add QnA Pairs for your bot
- Click on Save and Train to save the newly added QnA pair
Test and Publish the Knowledge Base
- Click on Test on top-right corner of the portal
- Enter a question and validate the response.
- Click on Test to close the panel
- Select Publish. Then to confirm, select Publish on the page.
Create a QnA Bot
- Click on Create Bot
- All details are pre-populated. Please change details as needed and click on Create
- After the bot is created, open the Bot service resource.
- Under Settings, select Test in Web Chat.
- At the chat prompt of Type your question, and enter:
QnA Maker supports single language per resource, and is explicitly set and determined from the files and URLs added when the first knowledge base of the service is created. This language can’t be changed for any other knowledge bases in the service.
I would encourage you to explore its other capabilities like multi-turn conversations and using pre-built API to query over text.