CLASSES2TEST
收藏资源简介:
CLASSES2TEST是由巴里理工大学与Wideverse公司联合创建的Java单元测试基准数据集,扩展自METHODS2TEST数据集。该数据集包含9,410个GitHub仓库的类级测试映射,涵盖真实项目中的完整类测试场景。数据集通过自动化流程从开源项目提取被测类与对应测试类的关联关系,支持类级交互测试的评估。主要应用于软件测试领域,旨在解决大语言模型在复杂类级别单元测试生成中的性能评估问题,为测试质量比较提供标准化基准。
CLASSES2TEST is a Java unit test benchmark dataset jointly developed by Polytechnic University of Bari and Wideverse Inc., which is extended from the METHODS2TEST dataset. This dataset contains class-level test mappings from 9,410 GitHub repositories, covering complete class-level test scenarios in real-world projects. The dataset extracts the association relationships between classes under test and their corresponding test classes from open-source projects via an automated pipeline, supporting the evaluation of class-level interactive testing. It is mainly applied in the field of software testing, aiming to address the performance evaluation issue of Large Language Models (LLMs) when generating complex class-level unit tests, and providing a standardized benchmark for test quality comparison.




