This repository contains the research artifact for the study "Why Do Developers Engage with ChatGPT in Issue-Tracker? Investigating Usage and Reliance on ChatGPT-Generated Code".
It includes the DevGPTIPlus dataset, scripts for data retrieval (GitHub Issues), and findings related to developers' usage and reliance on ChatGPT-generated code.
The data folder contains the dataset and results:
dataset/DevGPTIPlus.json: Contains the DevGPTIPlus dataset in JSON format.commits/: Contains all the issues with commits and their corresponding GPT-generated code (if any).
The survey_results folder contains the Certificate of Approval, thematic analysis, quantative results, and survey questions.
The src folder contains all the code used to generate the data:
data_collection/: Contains the data retrieval from GitHub REST API and filtering.rq1_usage/: Contains the implementation for RQ1.rq2_topic/: Contains the implementation for RQ2.rq3_code_reliance/: Contains the implementation code for RQ3.rq4_sentiment/: Contains the implementation code for RQ4.
Inside the Data folder
- RQ1: Contains the results of our open coding.
- RQ2: Contains the results of BerTopic modeling, including topics, key representations, and relevant documents, which we manually analyzed to determine the major categories.
- RQ3: Contains the results of XML files generated by CPD, which we manually analyzed.
- RQ4: Contains all the sentiment analysis for overall, usage-based and topic-based.
If you use this codebase, or otherwise found our work valuable, please cite:
@INPROCEEDINGS{das2025developersengagechatgptissuetracker,
author={Das, Joy Krishan and Mondal, Saikat and Roy, Chanchal K.},
booktitle={2025 IEEE International Conference on Software Analysis, Evolution and Reengineering (SANER)},
title={Why Do Developers Engage with ChatGPT in Issue-Tracker? Investigating Usage and Reliance on ChatGPT-Generated Code},
year={2025},
pages={68-79},
doi={10.1109/SANER64311.2025.00015}}