How does Service Desk determine which KB articles (related articles) to display?

[Michael Farrell]

We have a knowledge base that we have created in Confluence that we have connected to a Jira Service Desk. How are the related articles that display selected? What determines relevancy? Is it a combination of KB article Title, Content, and Tags? If the results are weighted what determines the best matches?

[Brant Schroeder · Accepted solution]

@Michael Farrell 

There is a little AI that is used to help best determine the answer.  Initially, it looks at the title, content on the article, and labels.  If you associated labels with specific request types this will also help.  As individuals who are searching for answers mark the article as helpful or if the article is used to answer a question it will use the keywords in the search or the request to try and match the article in the future for specific keywords.  So over time, it should get better at answering questions with more relevant articles.  When starting out using keywords in the title and body of the KB is important as well as matching labels with request types.  All this will help as you get started.

[Michael Farrell]

Thank you, Brant! This is a helpful response.

[Leandro Cantero]

@Brant Schroeder ,

Does using labels help with KB article suggestions in the customer portal only or does it help agents see suggested kb articles in the issue view in a project as well?

[Brant Schroeder]

I believe in both but it can be used to basically act as a filter in the portal.  In the portal you can limit the KBs displayed on a specific request type by the labels. 

[Leandro Cantero]

Got it. When I check the customer portal, the articles displayed there look fine and are as expected. However, in the agent's view, it's showing all articles without filtering based on the label configured for the request type of the issue.