How to develop an Azure chatbot using Q&A Maker?
Conversational AI chatbots are currently the rage in the market. Take any website, app or mobile site, chances are highly likely that you will be interacting with a bot to get your questions answered.
Specifically, chatbots are highly beneficial and useful in fintech, and healthcare segments, where some of the most redundant and repeating questions can be more easily answered by chatbots than humans. Apart from answering questions, chatbots are also used to accomplish tasks such as booking tickets, scheduling appointments and more.
If you’re someone looking to build a chatbot for your service, this post is just for you. Nowadays chatbots are very popular. We can see them on any website. It helps us to see and hear like humans. The main objective of the Chatbot is to book sessions or appointments with any business like a Teacher, doctor, etc. It is very easy to integrate any platform like Microsoft Teams, WhatsApp, Skype, and many more.
Here we go for Azure bot service and QnA maker.
Developing AI-Powered Chatbots Using Microsoft Azure QnA Maker:
QnA maker helps users create a chatbot without the need to code. To do that, we need to create a Knowledge base, where you add questions and relevant answers. Once all knowledge databases are created, you can initiate a test using the Webchat tab.
To do this, you first need to create an Azure Free or Purchase account. Simply go to https://portal.azure.com/
In Microsoft Cognitive Services, it is part of Azure. The Create a Knowledge Base page is wizard-like and the first step is where we will connect to Azure.
As shown in the above image, you need to create a QnA maker Azure service and fill in all the required fields. Then Press the ‘Review + Create’ button and after some time, a QnA maker instance is created. (In our case, the QnA maker’s Azure service name is sales-qna).
Now, move on to the Resource group (in which we have created App services) and select sales-qna clicked on ‘QnA Maker Portal’ under the second section.
Now, we are on the Q&A maker page. Here, we can create a Knowledge base by following these steps:
- In Step 2: Select Azure QnA service.
- In Step 3: Enter a knowledge base name.
- In Step 4: Select any type of file in your required format.
- For that please refer (https://docs.microsoft.com/en-us/azure/cognitive-services/qnamaker/reference-document-format-guidelines) post.
- Also, we can select URLs that support FAQ and Support.
Step-5: Click on the ‘Create your KB’ button
As shown above, display all the questions and answer pairs. You can also add a few more questions and answer pairs using the ‘+’ sign.
Sometimes users cannot be expected to answer in a single word or sentence. In such cases, we can use Multi-turn conversation.
Here (https://github.com/Azure-Samples/cognitive-services-sample-data-files/blob/master/qna-maker/data-source-formats/multi-turn.docx ) you can download that sample file for Multi-turn conversation.
If you want to upload more files and URLs into KB, use the ‘Settings’ button, edit and update KB name and URLs and Files. After updating any value, you need to click on the ‘Save and train’ button to retain the saved changes. Also, we can manage Files and URLs and also Import-Export KB.
Besides, we can test KBs using different cases before publishing. In that popup, we can test Questions, answers and multi-turn conversations.
You will now be able to see the ‘Create Bot’ button. Using this button, you can create a web bot.
The bot is almost ready. Add a name for your app and ensure it is different from the name you’ve given to your QnA maker service.
Specify SDK language here. After creating a bot, you can download the code from Azure to the local development environment and make custom changes.
Select Test in Web Chat to test your bot.
Wrapping Up
There you have it! Your first chatbot. As you keep working on it, you will figure out ways to make interesting changes and additions to optimize your chatbot and make it more interactive. Let us know in the comments what other features you would like to implement in it.