Automatic detection of Knowledge base articles in tickets

[Jean Waucomont]

Hello, 

In order to build a knowledge base for our support agents to resolve issues faster, I am trying to understand how knowledge articles are chosen to be linked to tickets. 
It is not an option for us to link individual tickets to the knowledge articles, as this would result in thousands of tickets linked to single articles.
Is there some kind of intelligence analysing the tickets for keywords?

I have been looking into Jira automation to link knowledge articles to Jira Service Desk tickets based on their content and/or ticket labels. 
In Jira automation there doesn't seem to be a way to do this. I can only add a comment to the ticket containing a link to the knowledge article. 
Additionally as automation rules based on the description field (ticket content) are very heavy, and we already have hundreds of rules, we don't want to go down that road.


I imagine someone must have had this question before, but I could not find related articles in the forum. 

The instructions listed in https://support.atlassian.com/jira-service-management-cloud/docs/how-do-you-automatically-present-articles-to-customers-in-a-request/ 
would require setting up a new type for every case (hundreds) to restrict the display of articles.

Please advise. 

[Brant Schroeder · Accepted solution]

@Jean Waucomont 

Jira has AI that makes suggestions based on keywords and past user behavior.  If users found the KB helpful and it answered their question or an agent used the article as an answer Jira learns and bases the results the user and agent sees on that.  You can help the AI by adding a specific label(s) to request types.  This helps the AI focus on a narrow set of KBs to provide results out of. https://support.atlassian.com/jira-service-management-cloud/docs/how-do-you-automatically-present-articles-to-customers-in-a-request/

Other than providing labels I am not aware of any other things you can do to help recommend a KB.

[Jean Waucomont]

Hello @Brant Schroeder
I have a follow-up question. It seems that (as you mentioned), the article suggestions by type are only possible for request types in Jira project settings. 

My question did not pertain to portal requests. 
We do not need knowledge article suggestions for requests via Portal at this point, but mainly for incident management (Jira Service Desk) and internal technical tasks (2 separate Jira projects). 
For incident management, the suggested articles should be matches for the issues in the ticket, based on the content of the ticket (and sometimes summary of the ticket). 
For technical tasks, I am planning to link how-to articles providing a step by step manual. 
The end result would then be that whatever ticket our support agents open, the correct knowledge article would be linked to it already. 

Since you said this works by AI, I am currently in the process of creating a bunch of tickets and linking them to specific KM articles in the hope that correct articles will start to pop-up as suggestions. 

Do you have any further tips to get this working?

[Brant Schroeder]

Nope, that is a great plan.  It will key in on keywords in the summary and description so make sure that those keywords are also in the KB article.

[Jean Waucomont]

Hello Brant, 
Thanks for the reply. I was planning to create dummy articles for demo purposes first, but since AI is involved, I will now create articles containing actual instructions, for better matching. 🤞 Let's hope it works at some point to catch border cases as well. We aim to have a 100% match, even for rare cases.