se2p/code-readability-merged
收藏资源简介:
该数据集名为Java Code Readability Merged Dataset,包含了421个Java代码片段及其可读性评分。这些数据是从多个科学论文中汇总而来的,主要用于训练Java源代码的可读性分类器。数据集的结构包括两个主要特征:code_snippet(Java源代码片段)和score(可读性评分)。评分是基于五级Likert量表,1表示非常不可读,5表示非常可读,评分是所有参与者的平均分。数据集由Buse Raymond PL、Dorn Jonathan和Sclabrino Simone整理,并由Krodinger Lukas共享。数据集的创建目的是为了推进代码可读性分类的研究,并且是现有数据集的合并和标准化版本。数据集的来源包括Buse、Dorn和Scalabrino的研究论文。数据集的局限性在于其规模较小,且评分主要由计算机科学学生提供,可能无法代表所有Java程序员。
This dataset is named Java Code Readability Merged Dataset. It contains 421 Java code snippets along with their readability scores. The data is aggregated from multiple scientific papers and is primarily intended for training readability classifiers for Java source code. The dataset structure includes two core features: `code_snippet` (Java source code snippet) and `score` (readability score). The scores are based on a 5-point Likert scale, where 1 indicates extremely unreadable and 5 indicates extremely readable, and the final score is the average rating across all participants. This dataset was curated by Buse Raymond PL, Dorn Jonathan, and Sclabrino Simone, and shared by Krodinger Lukas. It was developed to advance research on code readability classification, and is a merged and standardized version of existing datasets. The sources of the dataset include the research papers authored by Buse, Dorn, and Scalabrino. The limitations of the dataset include its relatively small scale, and the fact that the scores were primarily provided by computer science students, which may not be representative of all Java programmers.
Java Code Readability Merged Dataset
数据集描述
- 语言: 英语
- 许可证: 未知
- 大小类别: n<1K
- 任务类别: 文本分类
- 标签: readability, code, source code, code readability, Java
- 特征:
code_snippet: 字符串score: 浮点数
数据集信息
- 特征:
code_snippet: 字符串score: 浮点数
- 分割:
train: 354539 字节, 421 个样本
- 下载大小: 139793 字节
- 数据集大小: 354539 字节
数据集详情
数据集描述
- 包含: 421 个 Java 代码片段及其可读性评分
- 来源: 多个科学论文 [1, 2, 3]
数据集结构
- 每个条目包含:
code_snippet: Java 源代码片段score: 可读性评分(1.0 非常不可读 到 5.0 非常可读)
数据集创建
数据收集和处理
- 数据预处理: 对每个代码片段的可读性评分进行平均,并从三个来源合并
- 源数据: Buse, Dorn 和 Scalabrino 的论文
偏差、风险和限制
- 数据集大小: 非常小
- 评分者: 主要是计算机科学学生,不代表一般 Java 程序员群体
推荐用途
- 用于训练: 小型 Java 代码可读性分类器
引用
bibtex @article{buse2009learning, title={Learning a metric for code readability}, author={Buse, Raymond PL and Weimer, Westley R}, journal={IEEE Transactions on software engineering}, volume={36}, number={4}, pages={546--558}, year={2009}, publisher={IEEE} }
@inproceedings{dorn2012general, title={A General Software Readability Model}, author={Jonathan Dorn}, year={2012}, url={https://api.semanticscholar.org/CorpusID:14098740} }
@article{scalabrino2018comprehensive, title={A comprehensive model for code readability}, author={Scalabrino, Simone and Linares-V{a}squez, Mario and Oliveto, Rocco and Poshyvanyk, Denys}, journal={Journal of Software: Evolution and Process}, volume={30}, number={6}, pages={e1958}, year={2018}, publisher={Wiley Online Library} }




