Detecting Code Clones: A review

TL;DR

This review paper discusses various methods for detecting code clones, emphasizing the importance of robustness, adaptability across languages, and the need for standardized evaluation frameworks to improve practical industrial applications.

Contribution

It provides a comprehensive overview of existing code clone detection techniques and highlights the challenges and future directions for making these tools more practical and standardized.

Findings

String matching, token comparison, and AST-based methods are primary approaches.

Current tools lack standardized evaluation frameworks.

More research needed for industrial applicability.

Abstract

Code clone detection is involved with detecting duplicated fragments of code within a code base. Detecting these clones is useful for maintenance operations which require editing the clones. The tools developed are expected to be robust enough to identify clones even when they have been modified, whilst preserving reasonable recall and precision rates. It is also expected that these tools be easily adaptable to different programming languages. The major approaches to this problem has involve the use of direct string matching, token comparison or comparison using abstract syntax trees. It is difficult to compare detection tools due to the absence of a standardized framework for measurement. More work should be done to make the existing tools useful for other practical/industrial purposes.