CCBERT: Self-Supervised Code Change Representation Learning

This is the implementation of "CCBERT: Self-Supervised Code Change Representation Learning" published in the research track in ICSME 2023.

Download Replication package

Please download it from this link: https://zenodo.org/record/8153733

Setup conda environment

$ conda create -n ccbert python=3.9
$ conda activate ccbert
$ pip install -r requirements.txt

Downstream tasks

JIT Defect prediction

$  cd sh
$  bash defect_jdt.sh
$  bash defect_platform.sh
$  bash defect_gerrit.sh

Patch Correctness Assessment

$  cd sh
$  bash patch_correctness.sh

Bug-fixing Commit Prediction

$  cd sh
$  bash bug_fix_commit.sh