When saving a knowledgebase, QnAMaker reformats the answer.
- Additional \n characters are stripped (\n\n\n becomes \n\n) - The unicode non-breaking space character (U+00A0) is added. This character throws off most markdown implementations, resulting in incorrect rendering in e.g. Microsoft Teams.

On the QnAMaker website, please don't show API keys immediately on-screen and give users a button to copy them. Similar to the Azure portal, we don't want to inadvertently dox ourselves when using the website

We have multiple knowledgebases in one cognitive service and want to retrieve all of them with one get api call.. currently this is not supported in qna Maker. Can we share multiple comma separated kbids and get the all data merged together in one shot. these will develop the performance as well and would be helpful to us.

https://myhubfaqpartners.cognitiveservices.azure.com/qnamaker/v4.0/knowledgebases/{76d451e3-c81f-4243-a0c3-b9c772cb6be7},{6d775b61-8e14-4ba6-9ffe-f7bc27adb78e}/Test/qna

Can you change the font color to black instead of that light gray on the QNAMaker home page? It can be really hard to make out the light color fonts. Or maybe give the user an option to see black fonts?

Besides the logical OR operator for the api and strictFilter there should be an option for an unStrictFilter meaning if it finds the metadata request then it will use that filter and if not it will bring back the default answer.

Currently our API we our building will work like I am describing because things happen in a general sense... Every call can't be ultra specific to the API that is not feasible. However, it would be great along with the logical OR operator to be included in the API

Currently I can't use the QnA service at my project because of a region restriction (I can't deploy applications in west us)

Such a powerfull tool wasted here because a minor restriction!

There is any rational behind this region limitation?

Can devops be implemented for QnA just like luis as explained in NLU.DevOps? https://github.com/microsoft/NLU.DevOps

When adding a follow-up promt to question url formating don´t work in teams-channel but it works in web-channel

Is there a way to add feedback function when a chat ends or there's a certain keyword to trigger the feedback prompt? (like "Thanks"). Then have the feedback feed into azure and then create a query to pull the feedback table and information.

We've faced this issue while we're creating QnaMaker Service on Azure portal into a resource group when we have a location restricted resource group policy on Azure.

We're looking for only hosting services in North Europe or West Europe and Qna maker throws an error because of it's trying to create "QnAMaker Cognitive Service" to be hosted in West US.

Due to subscription limitation we can not create a resource group supporting West US region, our subscription only requires Europe based regions.

Please can you give an ability to QnAMaker Cognitive Service to pick location as well?

We've faced this issue while we're creating QnaMaker Service on Azure portal into a resource group when we have a location restricted resource group policy on Azure.

We're looking for only hosting services in North Europe or West Europe and Qna maker throws an error because of it's trying to create "QnAMaker Cognitive Service" to be hosted in West US.

Due to subscription limitation we can not create a resource group supporting West US region, our subscription only requires Europe based regions.

Please can you give an ability to QnAMaker Cognitive Service to pick location as well?

Enable existing Topics from Virtual Agent (PVA) to be processed by QnA Maker.
We have 180 topics and now we are told we will focus on QnA maker :( Looks like a load of work!

Alternative
What is the best practice on moving PVA topics to QnA maker?

Is there a query to get the question the user asked, when the bot answered with the string 'DefaultCardNoMatchResponse' (in QnAMakerBaseDialog.cs)?

I know we are able to get ones where the bot wasn't able to answer the question, but I want one that shows the question when user selects "none of the above" when the cards show "Did you mean..."

I have a request from a customer to support the CSV format, not just JSON as the Content-Type of the REST API.
Any plans to support csv format in the future?

It will be perfect if it can understands the questions in whichever language, then find the answers from the KB and return with the 'asked' language, optionally combine the answers in default language.

For example, I have KB in English, if the ask is in Japanese, it can understand and find the answer from the KB and return the answer in Japanese and the English (bilingual answer).

This will help a lot for preparing the multilingual KB when the chatbot is globally deployed.

How does the pricing work out? Can I have 20 knowledge base, each with 20 managed document and the pricing is still 10$ with 3 transactions per second. Or the pricing is based on 1 knowledge base?

In order to run reporting and analytics on the Knowledgebase the creation and update timestamp is meaningful.
At least in the XLS download function this data should be exposed.

We need more characters for additional follow up prompt, currently it has only maximum 50 chars only.

I cannot Save and Train after editing Setting.