Yesterday the team at Mozilla announced that the company is receiving hundreds of bug reports and feature requests from Firefox users on a daily basis. The team noted that it’s important to get the bugs fixed as soon as possible for the smooth functioning of the systems. Also, the developers should quickly come to know that there is a bug in order to fix it. Bug triage, a process where tracker issues are screened and prioritised can be useful in such cases.
However, even when developers come to know that bugs exist in the system, it is still difficult for the developers to closely look at each bug. The team at Mozilla has been using Bugzilla since years now which is a web-based general-purpose bugtracker and testing tool that group the bugs by product. But product assignment or the grouping process was done manually by the developers so this process failed to scale. Now Mozilla is experimenting with Machine Learning to train systems to triage bugs.
It’s important to get the bugs in the eye of the right set of engineers, for which the team at Mozilla developed BugBug, a machine learning tool that assigns a product and component automatically for every new untriaged bug. By bringing the bugs into the radar of the triage owners, the team at Mozilla has made an effort towards decreasing the turnaround time to fix new issues.
Training the BugBug model
Mozilla has a large training set of data for this model which includes two decades worth of bugs that have been reviewed by Mozillians and assigned to products and components. The bug data can’t be used as-is and any change to the bug after triage would create trouble during operation. So the team at Mozilla rolled back the bug to the time it was originally filed. Out of 396 components, 225 components had more than 49 bugs filed in the past 2 years. During operation, the team performed the assignment when the model was confident enough of its decision and currently, the team is using a 60% confidence threshold.
Ever since the team has deployed BugBug in production at the end of February 2019, they have triaged around 350 bugs. The median time for any developer to act on triaged bugs is 2 days. Usually, 9 days is the average time to act, but with BugBug the Mozilla team took just 4 days to remove the outliers.
Mozilla plans to use Machine learning in the future
The Mozilla team has planned to use machine learning to assist in other software development processes, such as identifying duplicate bugs, providing automated help to developers, and detecting the bugs important for a Firefox release. The team plans to extend BugBug to automatically assign components for other Mozilla products.
To know more about this news, check out the post by Mozilla.
*** This is a Security Bloggers Network syndicated blog from Security News – Packt Hub authored by Amrata Joshi. Read the original post at: https://hub.packtpub.com/mozilla-developers-have-built-bugbug-which-uses-machine-learning-to-triage-firefox-bugs/