Leveraging LLMs for Curating High-Quality~Bug~Report Datasets
收藏数据链接：
官方服务：
资源简介：
This repository contains (1) a manually labeled set of 2,000 bug reports and (2) five curated subsets from <b>QualDataBug</b> (SS1–SS5), selected using combinations of LLM classifications to reduce noise. The original BugHub dataset had ~40% noise, while the curated subsets have significantly lower noise levels: SS1 (9.58%), SS2 (16.25%), SS3 (19.29%), SS4 (21.59%), and SS5 (27.25%).
提供机构：
Anonymous_author创建时间：
2025-05-30
