This project would focus on using a Machine Learning algorithm to predict if a certain commit message is a bug fix or maybe a new immplemention. There is a naming convention that lots of commiters use when committing a bug fix or a new feature.
As such, the project train on the commit messages that follow this naming convention. then try to predict on that data if other commits that don't follow that naming convention are bug fixes or new implementations.
Currently, there are 1.6B commit messages saved on the DA servers so data collection should be relatively simple. The bulk of the work will likely be spent on creating the machine learning algorithm.
This is not a final idea so if you are interested in this project or wish to provide feedback, please raise an issue or comment on the issue I have already raised.