IRC-CLVul: Cross-Programming-Language Vulnerability Detection with Intermediate Representations and Combined Features

The most severe problem in cross-programming languages is feature extraction due to different tokens programming languages. To solve this problem, we propose a cross-programming-language vulnerability detection method paper, IRC-CLVul, based on intermediate representation and combined features. Specifically, first converted programs into unified LLVM (LLVM-IR) provide classification basis for Afterwards, extracted the code sequences control flow graphs of samples, used semantic model extract program information graph structure information, concatenated them vectors. Finally, Random Forest learn vectors obtained results. We conducted experiments 85,811 samples from Juliet test suite C, C++, Java. results show that our improved accuracy by 7% compared with two baseline algorithms, F1 score showed 12% increase.