JOSSE: A Software Development Effort Dataset Annotated with Expert Estimates

The JIRA Open-Source Software Effort (JOSSE) dataset consists of software development and maintenance tasks collected from the JIRA issue tracking system for Apache, JBoss, And Spring open-source projects. All the issues were annotated with actual effort and 19% of them were annotated with expert estimates. JOSSE is a task-based dataset with a textual attribute represented as a task description for each data point. This paper explains how the data were collected and details six data quality refinement procedures of the data points.

JIRA开源软件工作量数据集（JIRA Open-Source Software Effort, JOSSE）采集自Apache、JBoss与Spring开源项目的JIRA问题追踪系统，涵盖软件开发与维护任务。所有问题均已标注实际工作量，其中19%的问题还附带了专家估算值。JOSSE属于基于任务的数据集，每条数据点均包含以任务描述形式呈现的文本属性。本文详细阐述了该数据集的采集流程，并详述了针对数据点的六项数据质量优化流程。