Towards Reliable Code Plagiarism Detection: A Survey on Software Clone Detection
Sanjay B. Ankali, Shantappa G. Gollagi, Bahubali Akiwate · Zenodo (CERN European Organization for Nuclear Research) · 2023
Despite substantial study over the past three decades resulting in the development of more than 250 clone detection technologies, there is no one framework that can accurately and reliably identify all four major types of clones. The lack of comprehensive, reliable, and language-neutral code clone detection has a significant negative influence on online learning systems like Coursera, which are unable to assess the proficiency of students in coding projects and assignments they submit to the online platforms. This survey paper can contribute to building more reliable code plagiarism detection by presenting various tools and techniques to find the same language and cross-language clone types with respect to the clone types they detect and the languages they work on. The paper highlights 3 major issues in terms of language agnostic nature and accuracy a) Most of the proposed techniques work only on a specific language like C, CPP, Java, or Python for detecting clones. b) Only 8 proposed works accurately classify all 4 basic clone types. c) 98% of the clone detection in the past is based on regular clones ignoring micro clones. The summary of the paper can provide proper directions in building a more reliable code plagiarism detection tool.