Data for Common Sense-Violating Bugs Empirical Study

Research Context

Paper Title: An Empirical Study on Common Sense-Violating Bugs in Mobile Apps

Authors: Fu Fan, Yanjie Jiang, Tianyi Chen, Hengshun Zhang, Yuxia Zhang, Nan Niu, Hui Liu

Publication: ACM Transactions on Software Engineering and Methodology

Data Description

We release these parts of data from our empirical study:

The automatically collected issue dataset

• File: collected_issues.json

• The dataset consists of 33,650 issue reports retrieved from open-source Android applications.

• The JSON data is structured as a list of entries, where each entry includes the issue report's URL, title, body, and additional metadata.

The bug reports analyzed manually in our study

• File: analyzed_issues.csv

• This list contains 5,342 bug reports we have manually analyzed.

• The column "label" could be:

  • "Invalid" (excluded, not a valid bug report);

  • "Negative" (the bug do not violate common sense);

  • "Positive" (common sense-violating bug).

• The first 130 issues are selected from Andror2+ dataset with the same filtering strategy when constructing our dataset.

The common sense principles violated by the bugs with taxonomy

• File: common-sense-principles.json

• 358 common sense principles concluded from our study. Each principle is categorized by its level-1 and level-2 categories and includes links to the corresponding violating bugs.

Application Error Message Classification Dataset

• File: error_messages_dataset.csv

• This dataset includes 202 text samples with labels, some of which are derived from real-world bug reports.

The issue reports were retrieved in November 2022 using the GitHub API. The remaining data were generated through manual analysis.

Last Update: 2024-12-10