Exploring Architectural Design Decisions in Mailing Lists and their Traceability to Issue Trackers

The online repository for the paper: Exploring Architectural Design Decisions in Mailing Lists and their Traceability to Issue Trackers, published in ECSA 2024.

The online repo has the following folders:

1) Classifier: it contains the source code of the classifier and quantitative analysis. Furthermore, there are sufficient details on the classifier performance, and how to replicate the results.

2) Dataset: it contains the datasets we used to perform our analysis. This involves dataset before and after BERT classification, as well as exported json files used for training and analysis. The dataset in a zip file. This is a MySQL database in a zip file. It can be opened separately or opened using the search tool.

3) Qualitative analysis: it contains coding book of design decisions in mailing lists as well as coding book of methods to discuss ADDs between emails and issues, as well as precision charts for the applied similarity algorithms.

4) Searching tool: it contains the jar file and source code of the searching tool. The tool used to annotate and search for emails. It can be used to open the dataset from the zip directly. The tool is a jar file which can be run directly through a double click. The tool is tested on Windows and Linux. The folder also contains keywords to search for architectural emails, as well as documentation.

本数据集对应发表于ECSA 2024会议的论文《探索邮件列表中的架构设计决策及其与问题跟踪器的可追溯性》的在线开源仓库。
该在线仓库包含以下文件夹：
1) 分类器（Classifier）：该文件夹包含分类器的源代码与定量分析代码，详细阐述了分类器的性能表现，以及复现实验结果的具体方法。
2) 数据集（Dataset）：该文件夹包含本研究用于分析的全部数据集，涵盖BERT分类前后的数据集，以及用于模型训练与分析的导出JSON文件。本数据集为MySQL数据库压缩包，可直接解压打开，或通过配套检索工具加载使用。
3) 定性分析（Qualitative analysis）：该文件夹包含邮件列表中架构设计决策的编码手册，以及用于梳理邮件与问题间架构设计决策讨论方法的编码手册，同时收录了所使用相似度算法的精确率图表。
4) 检索工具（Searching tool）：该文件夹包含检索工具的JAR文件与源代码。本工具用于邮件标注与检索，可直接加载压缩包内的数据集；工具为可执行JAR文件，双击即可运行，已在Windows与Linux系统上完成测试。文件夹中同时收录了架构相关邮件的检索关键词与工具使用文档。