CSSG: Measuring Code Similarity with Semantic Graphs
Yiyang Lu, Jingwen Xu, Changze Lv, Zisu Huang, Zhengkang Guo · 2026
Read CSSG: Measuring Code Similarity with Semantic Graphs by Yiyang Lu, Jingwen Xu, Changze Lv, Zisu Huang, Zhengkang Guo (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-04085.txt.
Description
Yiyang Lu heads a team of nine authors proposing a new way to score how alike two programs are. Common scores lean on matching text or on syntax trees, so they miss deeper meaning. Two programs can look different yet do the same work, or look alike yet behave differently.
Their metric, CSSG, builds semantic graphs from program dependence graphs, which record control links and how variables affect each other. On the CodeContests+ data set, CSSG beats earlier metrics at telling more alike code from less alike code. The advantage holds both within one language and across languages. Lu and his eight co-authors argue that dependence-aware graphs give a stronger base for similarity than surface text or syntax alone.
Catalog record
| CREATOR | Yiyang Lu, Jingwen Xu, Changze Lv, Zisu Huang, Zhengkang Guo |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 19 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-04085