Paper accepted at IEEE TSE

Paper titled " Modeling Functional Similarity in Source Code with Graph-Based Siamese Networks", authored by Nikita Mehrotra (PhD student, IIIT Delhi), Navdha Agarwal (B. Tech. IIITD), Piyush Gupta (B. Tech. IIITD), Saket Anand (Associate Professor, IIIT Delhi ), David Lo (Professor, SMU Singapore) and Rahul Purandare (Associate Professor, IIIT Delhi), has been accepted as a Journal First for publication in the upcoming issue of the Transactions on Software Engineering (TSE).

This work is supported in part by the Department of Scienceand  Technology  (DST) India,  Science  and  Engineering Research Board (SERB), the Confederation of Indian Industry(CII), Infosys Center for Artificial Intelligence at IIIT-Delhi, and Nucleus Software Exports Ltd.

Abstract: The paper addresses the problem of semantic code clone detection. Semantic clones are duplicate code fragments that share similar semantics but are different syntactically. Code clone detection plays an important role in software maintenance, code refactoring, and reuse. A substantial amount of research has been conducted in the past to detect clones. A majority of these approaches detect syntactic clones, and only a few of them target semantic clones. This paper addresses the problem of semantic code clone detection using program dependency graphs and geometric neural networks. The authors have developed a prototype tool HOLMES, based the proposed approach and empirically evaluated it on popular code clone benchmarks. The results show that HOLMES performs considerably better than the other state-of-the-art tool, TBCCD. The unseen projects and cross dataset experiments show that HOLMES outperforms TBCCD and most of the pairs that HOLMES detected were either undetected or suboptimally reported by TBCCD.