Replication Package for the Paper: "Understanding Code Smell Detection via Code Review: A Study of the OpenStack Community"

This repository contains the data and results from the paper "Understanding Code Smell Detection via Code Review: A Study of the OpenStack Community" submitted to ICPC 2021.

1. "data.zip" contains the following three folders:

1) data folder

The data folder contains the retrieved 1,190 reviews that discuss code smells. Each review includes four parts: Code Change URL, Code Smell, Code Smell Discussion, and Source Code URL.

2) scripts folder

The scripts folder contains the Python scripts that were used to search for code smell terms and the list of code smell terms.

keyword.txt contains the keywords associated with code smells, such as "smell, duplication, and dead".

get_changes.py is used for getting code changes from OpenStack.

get_comments.py is used for getting review comments for each code change.

keywords_search.py is used for searching review comments that contain at least one keyword.

random_select.py is used for randomly selecting review comments that do not contain any keyword.

keywords_improve.py is used for improving the keyword-based mining approach.

tools.py is used for supporting the process of keywords improving.

3) project folder

The project folder contains the MAXQDA project files. The files can be opened by MAXQDA 12 or higher versions, which are available at https://www.maxqda.com/ for download. You may also use the free 14-day trial version of MAXQDA 2018, which is available at https://www.maxqda.com/trial for download.

Data Labeling & Encoding for RQ2.mx12 is the results of data labeling and encoding for RQ2, which were analyzed by the MAXQDA tool.

Data Labeling & Encoding for RQ3.mx12 is the results of data labeling and encoding for RQ3, which were analyzed by the MAXQDA tool.

2. Keywords associated with code smells.pdf

This file contains the final set of keywords associated with code smells that we identified by following the systematic approach proposed by Bosu and his colleagues in their paper: Identifying the Characteristics of Vulnerable Code Changes: An Empirical Study, FSE 2014.