An Azure Cognitive Service for Language feature that distills information into conversational, easy-to-navigate answers
Quickly create a conversational layer over your data
Create a conversational question-and-answer layer over your existing data with question answering, an Azure Cognitive Service for Language feature. Build a knowledge base by adding unstructured documents or extracting questions and answers from your semi-structured content, including FAQ, manuals, and documents. Get the best answers from the questions and answers in your knowledge base—automatically. Your knowledge base gets smarter, too, as it continually learns from user behavior.
Automatic question-answer extraction from semi-structured content, including FAQ, product manuals, guidelines, support documents, and policies
Knowledge base with suggestions for alternative questions. Add or reject them based on their relevance
Complex multiturn conversations can be easily created, edited, and trained in the new language studio or REST APIs
Bots can be created and published without code experience. Simply upload a semi-structured document or URL
Add personality to your bot using chit-chat
Make your bot more conversational by adding a pre-populated chit-chat dataset. Choose a persona that’s professional, witty, friendly, caring, or enthusiastic.
Support rich data and dynamically manage content
Add rich content to your knowledge base with built-in rich text editor. Use Markdown’s full capability to author answers directly or upload a knowledge base that includes rich content.
Scale to meet your business and compliance needs
Question answering is a hosted model. Choose product tiers according to your size and throughput needs, and feel confident knowing all components are within your Azure compliance boundary.
Chat in more than 50 languages
Select the language for your question answering use case or opt to create knowledge bases in multiple languages when you create the first knowledge base.
Microsoft invests more than USD 1 billion annually on cybersecurity research and development.
We employ more than 3,500 security experts dedicated to data security and privacy.
Azure has more certifications than any other cloud provider. View the comprehensive list.
Create a feedback loop
Use active learning to improve the quality of your knowledge base. Generate alternative questions based on user submissions. Review those suggestions and add or reject them based on their relevance.
Everything you need to get started
See how trusted companies are using question answering, a feature of Language service
Accenture eases the way for new hires
Sangita Agarwal Managing Director, Human Resources Technology Platforms, Accenture
"QnA Maker enabled Accenture to build corpus at speed to quickly launch an HR chatbot for 100,000-plus people. We were able to leverage QnA Maker and LUIS to build our conversation hierarchy, getting us the best of both."
Spanish telco builds digital assistant
Antonio Guzman Head of Innovation, Aura, Telefonica
"Using Bot Service and Cognitive Services, such as Language Understanding and the Azure QnA Maker API, made it easy for our developers to put an advanced, conversational experience in the hands of our customers very fast."
Insurance shoppers engage with AI chatbot
Matt White Marketing Manager, Personal Lines Acquisition Experience, Progressive Insurance
"By using Azure Bot Service and Cognitive Services, such as Language Understanding and QnA Maker, we've been able to continue our own Progressive journey of digital innovation, and do it in an agile, fast, and cost-effective way."
Frequently asked questions about question answering
Question answering introduces several new features including enhanced relevance using a deep learning ranker, support for unstructured documents as a data source, ability to generate precise answers, and end-to-end region support.
All delete operations within question answering are permanent, including question-and-answer pairs, files, URLs, knowledge bases, and Azure resources. Be certain to export your knowledge base before deleting any part of it.
Follow the instructions in the documentation to create a bot with question answering.
Question answering supports multiple languages. Create a knowledge base in multiple languages in a single language resource.