Filtering of commits - Githubissues

fizzerle / Commit-Mentor

This Projects helps developers to write better commit messages.

MIT License

0 stars 0 forks source link

Filtering of commits #3

Closed fizzerle closed 3 years ago

fizzerle commented 3 years ago

Commits like merge and auto generated commits by bots should be automatically removed

fizzerle commented 3 years ago

Remove Commits that are:

empty
contain non-ASCII messages
are not written in english (maybe irrelevant because the projects that are chosen are already in english)

Also report how many commits that are removed. This is interesting, because we can evaluate if hypothesis that we extract good commit messages by taking the commit messages by the most active developers holds.

Paper that uses these criterias:

Zeitschriftenaufsatz Liu, Shangqing; Gao, Cuiyun; Chen, Sen; Lun Yiu, Nie; Liu, Yang (2020): ATOM: Commit Message Generation Based on Abstract Syntax Tree and Hybrid Ranking. In: IEEE Transactions on Software Engineering, S. 1. DOI: 10.1109/tse.2020.3038681.

fizzerle commented 3 years ago

This is the Page that describes what jiang et al. did to the data: https://sjiang1.github.io/commitact/index.html

In short they

removed the merges and rollbacks from our data set. To detect the mereges and rollbacks, we check whether the messages are started with "merge" or "rollback"

fizzerle commented 3 years ago

Regex used by Liu et al. cleaned dataset of jiang:

message_patterns.pdf

Beitrag Liu, Zhongxin; Xia, Xin; Hassan, Ahmed E.; Lo, David; Xing, Zhenchang; Wang, Xinyu (2018): Neural-machine-translation-based commit message generation: how far are we? In: Marianne Huchard, Christian Kästner und Gordon Fraser (Hg.): Proceedings of the 33rd ACM/IEEE International Conference on Automated Software Engineering. ASE '18: 33rd ACM/IEEE International Conference on Automated Software Engineering. Montpellier France, 03 09 2018 07 09 2018. New York, NY, USA: ACM, S. 373–384. Schlagwörter: Commits message generation Gruppen: Paper