BigIssue

BigIssue是由Salesforce Research创建的现实Bug定位基准数据集，包含10,905个来自4,233个Java仓库的真实Bug和合成Bug。该数据集通过GitHub Issues系统报告并由提交或拉取请求修复。此外，还提供了一个长序列合成Bug数据集，由先进的代码生成模型InCoder生成。BigIssue旨在通过提供多样化的Bug数据，推动Bug定位技术的进步和自动程序修复性能的提升，以适应现代开发周期。

BigIssue is a real-world bug localization benchmark dataset developed by Salesforce Research, which contains 10,905 real and synthetic bugs sourced from 4,233 Java repositories. All these bugs are reported via the GitHub Issues system and fixed through commits or pull requests. Additionally, a long-sequence synthetic bug dataset generated by the state-of-the-art code generation model InCoder is also provided. BigIssue aims to advance bug localization technologies and improve the performance of automated program repair by offering diverse bug data, so as to cater to modern software development cycles.

搜集汇总

背景与挑战

背景概述

BigIssue是一个现实Bug定位基准数据集，包含10,905个来自4,233个Java仓库的真实和合成Bug，旨在推动Bug定位技术和自动程序修复性能的提升。