Recommending Code Understandability Improvements based on Code Reviews

TL;DR

This paper proposes a system that leverages code review data and machine learning to recommend improvements for code understandability, aiming to reduce developer effort and bugs.

Contribution

It introduces a novel dataset of understandability improvements from code reviews to train ML models for recommending enhancements.

Findings

Dataset of code review-based understandability improvements created

Initial framework for ML-based recommendation system proposed

Potential to improve code readability and reduce bugs

Abstract

Developers spend 70% of their time understanding code. Code that is easy to read can save time, while hard-to-read code can lead to the introduction of bugs. However, it is difficult to establish what makes code more understandable. Although there are guides and directives on improving code understandability, in some contexts, these practices can have a detrimental effect. Practical software development projects often employ code review to improve code quality, including understandability. Reviewers are often senior developers who have contributed extensively to projects and have an in-depth understanding of the impacts of different solutions on code understandability. This paper is an early research proposal to recommend code understandability improvements based on code reviewer knowledge. The core of the proposal comprises a dataset of code understandability improvements extracted from code reviews. This dataset will serve as a basis to train machine learning systems to recommend understandability improvements.