OSSScope builds a code-clone feature set from 190K C repos
Original title: Effective Construction of Feature Database of Fingerprinting Functions for Code-Clone SCA
The selected set holds 61 million fingerprinting functions and is reported to improve recall 22.65% over prior tools.
Researchers propose OSSScope, a greedy method for choosing GitHub repositories so code-clone detection databases cover more unique fingerprinting functions. Applied to 2.55 million C repositories, it selected 190K repositories and 61 million functions, covering over 95% of features from other sources. They also introduce RClassifier to decide whether a repository should be included when moving the approach to other languages. On Golang, its recall was comparable to the OSSScope-built dataset.
Unconfirmed: The 22.65% recall improvement is the authors' comparison with state-of-the-art tools.
- OSSScope analyzed fingerprint diversity across 2.55 million C GitHub repositories.
- The method aims to maximize unique functions while minimizing repository count.
- The selected features cover over 95% of features from other sources.
- RClassifier is meant to streamline migration to other language ecosystems.
- Products
- OSSScopeRClassifier
- Organizations
- GitHub
From the original article
1 min read at arxiv.orgIn contemporary software development, the extensive adoption of third-party libraries has become standard practice to avoid reinventing the wheel. This trend is particularly evident in programming ecosystems that lack mature package managers, such as C/C++.
By Zhiling Zhu · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.