Replication Package for "Beyond the Good First Issue: Progressive Recommendation of Second Tasks for Newcomers in OSS Communities"

This repository contains the data and scripts used in the paper: "Beyond the Good First Issue: Progressive Recommendation of Second Tasks for Newcomers in OSS Communities" Data-Preparation The folder Data-Preparation/ includes the following resources: NewcomerList.csv A list of identified newcomers based on the public GFI-Bot dataset [1]. PRInfo/, PRComments/, issueInfo/, IssueComments/, commitInfo/ Activity data collected via the GitHub API for repositories involving these newcomers, including pull requests, issues, comments, and commits. get_issue_chain.py Identifies issues resolved by newcomers and extracts resolver information. Outputs issue resolution data to issueGetResolver/ Constructs newcomer task chains and stores them in IssueChainGE2/ RQ1 The folder RQ1/ contains scripts for feature extraction and statistical analysis: cal_issue_feature.py Extracts and enriches issue features used in the RQ1 analysis. similarity-compare.py Computes feature similarity between issues and performs Wilcoxon tests for issue-level similarity analysis. Output: RQ1/Result/similarity_wilcoxon_results.csv progression-compare.py Evaluates issue-level progression patterns using Wilcoxon tests. Output: RQ1/Result/progression_wilcoxon_results.csv network-compare.py Gets social network metrics and performs Wilcoxon tests for social network analysis. Output: RQ1/Result/network_wilcoxon.csv RQ2 The folder RQ2/ contains scripts for dataset construction and model evaluation: 0-get_first_participate_time.py Identifies the timestamp of a newcomer’s first participation in the second issue (recommendation time). 1-get_candidate_list.py Constructs candidate issue lists for each newcomer. 2-get_candidate_issue_feature.py Extracts features for each (first issue, candidate issue) pair, including: Attributes of the first issue (FI) Attributes of candidate issues Similarity features between FI and candidate issues Output stored in CandidateIssueInfo/ 3-get_rec_dataset.py Constructs the dataset for training and evaluating recommendation models. 4-recommend.py Trains learning-to-rank models and performs evaluation and ablation studies. Reference [1] He H, Su H, Xiao W, et al. GFI-Bot: Automated Good First Issue Recommendation on GitHub Proceedings of the 30th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE), 2022, pp. 1751–1755.