Dependabot and Security Pull Requests

This deposit contains three (3) datasets that were used in the study "<em>Dependabot and Security Pull Requests: Large Empirical Study</em>" (Under review). Each dataset is described as follows : <strong>Dataset (1) - Dependency Update</strong> : This dataset concerns issues related to pull requests that were created by both users and bots to manage dependency updates in GitHub projects. The search query that we used consists of looking for the keywords "<em>Dependency, Update</em>" in the title, body or comment of a PR created in the time period between 26/05/2017 and 15/06/2021. Note that the first date corresponds to the launch date of Dependabot that we chose as reference for all our datasets to conserve homogeneity and uniformity of PRs creation dates, and the second date corresponds to the day that we executed our collection pipeline process. We obtained a total of 6,573,489 PR-related issues belonging to a total of 927,007 repositories. <strong>Dataset (2) - Dependabot Security PRs</strong> : The second dataset is related to PRs created by Dependabot to handle security vulnerabilities in project dependencies. In the search query, we look for PR-related issues created by "<em>Dependabot-preview</em>" and with the label "<em>security</em>", also created during the time period between 26/05/2017 and 15/06/2021. With these parameters, the GitHub API returned 384,764 issues from 36,911 repositories. <strong>Dataset (3) - Manual Security PRs</strong> : For this dataset, we were interested in PRs created only by users to handle security vulnerabilities. The used query consists of searching for the keywords "<em>Dependency, Vulnerable</em>" in the title, body or comment of a PR created in the time period between 26/05/2017 and 15/06/2021. After the pipeline execution, we excluded the issues that were created by bots by filtering the PR authors with the GitHub user type "<em>bot</em>" and only keeping those with the type "<em>user</em>". The final results include a total of 100,102 issues for 37,845 repositories. It is worth noting that we have explored several search keywords and considered using root-based search to make sure that we covered the maximum amount of cases without missing any PRs.