Introducing Question Answering in Public Preview

This article is contributed. See the original author and article here.

QnA Maker is an Azure Cognitive Service that allows you to create a conversational layer over your data in minutes. As part of our AI at Scale initiative across Microsoft, we are making the latest breakthroughs in natural language understanding available across our products and within our Azure Services portfolio. Powered by our Turing natural language model, we are excited to introduce the new Custom Question Answering feature (public preview) with Text Analytics. Alongside Custom question answering, we are also introducing a new Prebuilt Question answering capability that lets users extract relevant answers to questions from a given passage of text.

Overview

Custom question answering supports all capabilities introduced with QnA Maker managed.

You can add unstructured files to your knowledge bases with Custom question answering.

We are also introducing Prebuilt question answering capability that lets users extract relevant answers to questions from a given passage of text.

Enabling Custom question answering feature

As Custom question answering is now a feature in Text Analytics, users should first visit the Text Analytics resource create blade on Azure portal. They will have the option to enable Custom question answering feature within the resource. Once the Custom question answering feature is selected, the user will update all details related to the feature in the create blade. On creation, a new Text Analytics resource with Custom question answering feature is deployed.

If users don’t enable the feature on resource creation, they will have the option to enable the feature later through the Features tab of the Text Analytics resource blade. They will also be able to disable the Custom question answering feature via the Features tab.

Support for unstructured documents

With Custom question answering (preview), we are introducing the ability to add unstructured files to the knowledgebases. Now the content author can ingest entire documents in the knowledgebase and when a user query is passed, a response is returned on searching the documents ingested. The content authors don’t need to manage QnA pairs for unstructured documents.

Take for instance the following unstructured document introducing Surface Products: SurfaceBlog. The user can ingest this document in the knowledge base and test for user queries from the text. We tested this with a few queries as shown below.

Prebuilt Question Answering

Prebuilt question answering provides users the capability to answer questions over a passage of text without having to create knowledge bases and manage additional storage. This functionality is provided as an API and can be used without having to learn the details about QnA Maker. Given a user query and a block of text/passage the API will return an answer and precise answer (if available). We currently support the ability to pass text from 5 documents.

API Updates

We have added a new preview release for v5.0 APIs: V5.0-preview.2. Users should access this version for unstructured documents support and pre–built question answering.

Pricing

Custom question answering feature is free in public preview. However, users will be charged for Azure cognitive search they add to the feature as per the tier selected.

Note to existing QnA Maker managed users

QnA Maker managed has been re-introduced as Custom question answering feature in Text Analytics.

Going forward users will not be able to create a new QnA Maker managed resource. All new resources can be created by enabling Custom question answering feature with Text Analytics.

All existing QnA Maker managed (preview) resources continue to work as before. The resource settings, portal, endpoints, keys, SDK, etc. pertaining to existing QnA Maker managed resources will work as is. There is no action required from the user at this point in time.

We will continue to support v5.0-preview.1 for existing QnA Maker managed customers.

If the users choose to migrate to Custom Question answering, they can do so. To migrate from QnA Maker managed to Custom question answering, users can create a Text Analytics resource with Custom question answering feature enabled and migrate the knowledge bases from QnA Maker managed to the Text Analytics resource.