How we use machine learning in the Embold Assignee Suggestion on Jira?

Here are few points which explain the problem, the solution we have provided and the advantages -

Problem - Whenever a bug is reported in any issue tracking tool, it’s important to know whom the bug should be assigned to, or who is more likely to fix the bug. Usually, the bug is assigned to the wrong developer, which requires re-assigning and results in loss of expensive engineering work hours. Predominantly Managers or leads are the people to take up tasks to determine the rightful person whom the ticket should be assigned to. If this process could be automated, it would help save a lot of valuable time.

Solution - A machine learning based approach is introduced which learns from past data fetched from issue tracking tool like JIRA and extracts patterns between the information from issue ticket like Title, description etc. and the person who solved it in the past, the assignee. Then whenever a new ticket is filed, these patterns are then used to identify top candidates from that organization / project who have worked on similar issues earlier, and are most likely to solve this problem as well.

Advantages -

  • Saves expensive engineering hours of Leads / Managers and therefore reduces total operational cost of organization.
  • Very useful once the organization / project has expanded and it is hard for a human to keep track of who should be the rightful assignee.