QnA Maker is being retired. Hello, Question Answering!

This post has been republished via RSS; it originally appeared at: Microsoft Tech Community - Latest Blogs - .

At Microsoft Ignite, 3rd November 2021, we announced the new Question Answering feature within the Cognitive Service for Language. This capability is the newer and improved version of the erstwhile standalone QnA Maker. Following this launch, the QnA Maker service was put on a retirement path, with the deprecation date set as March 31, 2025.


This article aims to answer a few common questions around this announcement.


Q. What is Question Answering?

A. Powered by state-of-the-art transformer models and Turing natural language model, Question Answering is Microsoft Azure’s latest intelligent Q&A offering with marked improvements in relevance and quality over QnA Maker. For instance, while Question Answering has the same custom Q&A capabilities familiar in QnA Maker, it also introduces a new capability to query a text passage and get a response without having to manage knowledge bases, using prebuilt question answering models.


Q. Why is QnA Maker being retired?

A. QnA Maker became Generally Available in 2018. Since then, we have seen good engagement and feedback from our customers. However, given new advancements in computation capabilities and natural language models, the old QnA Maker service needs to be retired and replaced with a new architecture to bring you the best of Azure’s Language capabilities.


Question Answering will have a much-simplified architecture, which will remove the requirement to host a separate App Service, thus making the management of the service easier. The other benefit is that customers will only need to pay for the service and Azure Cognitive Search. For QnA Maker, customers had to pay an additional cost for Azure App Service. 


Q. When is the retirement happening?

A. QnA Maker is scheduled to retire on March 31, 2025.


Q. Can I use my QnA Maker resources till the retirement date?

A. Existing QnA Maker resources will be officially supported till March 31, 2025. However, customers will be unable to create new QnA Maker resources starting October 1, 2022.


Q. What happens to my existing QnA Maker knowledge bases?

A. Your existing QnA Maker knowledge bases will continue to work as before. You can create new knowledge bases on your existing QnA Maker resources until the retirement date. However, we strongly advise you to migrate these knowledge bases to the new Question Answering service at the earliest.


Q. I have further questions on retirement. How can I get assistance around it?

A. If you have questions, get answers from community experts in Microsoft Q&A. If you have a support plan and you need technical help, please create a support request:


  1. Enter a summary of your issue.
  2. Under Issue type, select Technical.
  3. Under Subscription, select your subscription.
  4. Under Service, select My services.
  5. Under Service type, select Cognitive Services.
  6. Under Problem type, select QnA Maker–Knowledgebase Management.
  7. Under Problem subtype, select Migration, import, export, knowledge base.

You can also reach out to the Question answering product team at qnaproductmanagers@microsoft.com.


Q. How do I access Question Answering?

A. At Microsoft Ignite 2021, we announced the unified Cognitive Service for Language, which brought existing Azure services for language understanding, text analytics, and question answering into one Azure resource. As a result, you can access question answering as a feature within the Language resource.


Q. What is new in Question Answering?

A. With Question answering, we are introducing:


  1. Access to Question Answering through the ‘answer questions’ section in the language studiothe recently announced customization portal for your Language service.
  2. Availability across 32 regions supported by the Language Service.  
  3. Support to add and query unstructured content from public URLs with custom question answering for 53 languages .
  4. Precise answering support for 53 languages.


Q. What capabilities does Question Answering provide?

A. Question answering comprises of two capabilities:


  1. Custom question answering: This enables users to customize different aspects of their question answering model, like the ability to edit question and answer pairs extracted from a content source, define synonyms and metadata, accept question suggestions, and more.
  2.  Prebuilt question answering: This new capability allows users to querying a text passage and get a response without having to manage knowledge bases.


Q. Why should I migrate my QnA Maker knowledge bases to Custom question answering?

A. As part of our AI at Scale initiative across Microsoft, we are making the latest breakthroughs in natural language processing available across our products and within our Azure Services portfolio. Powered by state-of-the-art transformer models and Turing natural language model, Custom question answering offers the latest and best functionalities with marked improvements in relevance and quality over QnA Maker. Additionally, all subsequent updates and technological improvements will be made on Custom question answering only.


Q. How do I migrate my knowledge bases to Custom question answering?

A. Please follow the steps listed in this document to migrate your knowledge bases: Migrate QnA Maker knowledge bases to custom question answering - Azure Cognitive Services | Microsoft Docs

The figure below displays the changes from the previous QnA Maker to the new Question answering service:





Q. How much will I be charged for using Question Answering?

A. Please find the pricing details for Question answering here: Pricing - Language Service | Microsoft Azure.


Q. I have heavy dependencies on the QnA Maker in my business. Can the retirement date be pushed?

A. The retirement date will not be pushed further than 31st March, 2025. Question Answering offers new and improved capabilities over QnA Maker and we strongly advise all users to migrate as soon as they are able.

Leave a Reply

Your email address will not be published. Required fields are marked *


This site uses Akismet to reduce spam. Learn how your comment data is processed.