license: mit
Programminglanguage: Java
version: N/A
Date: May 2019 paper release date for https://arxiv.org/pdf/1812.08693.pdf
Contaminated: Very Likely
Size: Standard Tokenizer
Dataset is imported from CodeXGLUE and pre-processed using their script.
Where to find in Semeru:
The dataset can be found at /nfs/semeru/semeru_datasets/code_xglue/code-to-code/code-refinement/data/small in Semeru
Task Definition
Code refinement aims to automatically fix bugs in the code, which can contribute to reducing the cost of bug-fixes for developers. In CodeXGLUE, given a piece of Java code with bugs, the task is to remove the bugs to output the refined code. Models are evaluated by BLEU scores, accuracy (exactly match) and CodeBLEU.
Dataset
We use the dataset released by this paper(https://arxiv.org/pdf/1812.08693.pdf). The source side is a Java function with bugs and the target side is the refined one. All the function and variable names are normalized. Their dataset contains two subsets ( i.e.small and medium) based on the function length. This dataset is small.
Data Statistics
Data statistics of this dataset are shown in the below table:
| #Examples | |
|---|---|
| Small | |
| Train | 46,680 |
| Valid | 5,835 |
| Test | 5,835 |
Reference
@article{tufano2019empirical,
title={An empirical study on learning bug-fixing patches in the wild via neural machine translation},
author={Tufano, Michele and Watson, Cody and Bavota, Gabriele and Penta, Massimiliano Di and White, Martin and Poshyvanyk, Denys},
journal={ACM Transactions on Software Engineering and Methodology (TOSEM)},
volume={28},
number={4},
pages={1--29},
year={2019},
publisher={ACM New York, NY, USA}
}