QnA Maker Knowledge Base returning ‘No good match found in KB’ for follow-up prompts

Question

I’ve got an issue with my QnA Knowledge Base where it’s retuning ‘No good match found in KB’ when using follow-up prompts that I've configured to return an answer.

I have c.200 question/answer pairs set up and all of them have follow-up prompts linking each question/answer pair to other question/answer pairs. However, when I test the knowledge base in QnA Maker I get an answer of ‘No answer found in KB’.

Below is an example:

I have a question/answer pair to answer the question ‘What is depression’ that has five different follow-up prompts - Prevalence, Causes, Types, Symptoms and Related Issues:

As you can see in the below image, the Prevalence follow-up prompt is configured to answer using the ‘How common is depression’ questions/answer pair:

However, when I test this using QnA Maker’s built in test chatbot I get the answer ‘No good match found in KB’:

When I inspect the result I see the following:

As you can see, there is no answer returned and the confidence score is 'None'.

Has anyone else seen this problem before and have a solution?

Matt Stannett Matt Stannett · Accepted Answer · 2019-07-10T09:18:52

Follow-up prompts aren't currently supported out of the box outside of the QnA Maker portal. There is a C# and NodeJS experimental samples available which demonstrate how you can integrate this functionality into your bot.

Since you haven't specified a language preference, I will go with the C# one, basically your QnA code needs to be updated from something like:

var qnaMaker = new QnAMaker(new QnAMakerEndpoint
{
    KnowledgeBaseId = _configuration["QnAKnowledgebaseId"],
    EndpointKey = _configuration["QnAEndpointKey"],
    Host = _configuration["QnAEndpointHostName"]
},
null,
httpClient);

var response = await qnaMaker.GetAnswersAsync(turnContext);

if (response != null && response.Length > 0)
{
    await turnContext.SendActivityAsync(MessageFactory.Text(response[0].Answer), cancellationToken);
}
else
{
    await turnContext.SendActivityAsync(MessageFactory.Text("No QnA Maker answers were found."), cancellationToken);
}

to:

var qnaMaker = new QnAMaker(new QnAMakerEndpoint
{
    KnowledgeBaseId = _configuration["QnAKnowledgebaseId"],
    EndpointKey = _configuration["QnAEndpointKey"],
    Host = _configuration["QnAEndpointHostName"]
},
null,
httpClient);

var response = await qnaMaker.GetAnswersAsync(turnContext);
var qnaAnswer = response[0].Answer;
var prompts = response[0].Context?.Prompts;

if (prompts == null || prompts.Length < 1)
{
    await turnContext.SendActivityAsync(MessageFactory.Text(response[0].Answer), cancellationToken);
}
else
{
    // Set bot state only if prompts are found in QnA result
    newState = new QnABotState
    {
        PreviousQnaId = response[0].Id,
        PreviousUserQuery = query
    };

    outputActivity = CardHelper.GetHeroCard(qnaAnswer, prompts);
}

The relevant code is in this file, as you can see there is some additional code involved to store the progress through the follow-up prompts, so perhaps it might be easier to plug in your KB details to the sample then try it out to see how it works before porting it over into your bot.

QnA Maker Knowledge Base returning ‘No good match found in KB’ for follow-up prompts

2 Answers