Problem Definition

Currently, the "Related Knowledge Base Articles" generates article suggestions that might be related to the ticket, it is not always accurate and doesn't tell why is the article even suggested. The user needs to manually check what matches and what doesn't. This leads to some articles which are empty being suggested too.

Suggested Solution

Improving the algorithm on how the articles match would be highly appreciated. Though this might be a tough undertaking.

In the interim, if there is a system that shows "matches" and highlights as to why it was shown as a "Related knowledge base article", that would be really helpful to know and understand the suggestion. 

For example, issue summary: "This is a test issue"
Article Suggested: How to Troubleshoot and perform Tests
Match highlights: "This article was chosen because the word "test" was matched. 

This provides some clarity and doesn't mislead the agent into thinking the article has some linking when it doesn't. 

Create the ability to add a list of words (like conjunctions) to exclude from matches.