Project structure

• AutoenCODE: This folder contains the source code of word2vec.
• graph2vec: This folder contains the source code of graph2vec.
• joern: This folder contains the source code of joern, which is used to extract AST and CFG.
• script: This folder contains the deep learning model to complete the analysis tasks used in our experiment.
• src: This folder contains the code of extracting the syntactic and semantic embedding features of the C++ source code.
• data: This folder contains the fusion feature data of three analysis tasks.

Usage

Generating the embedding features

The src/main/java/test/GenerateTrainingData.java file is the entrance of generating embedding feature data set for the C++ source code. The embedding features of the data set have been stored in the data folder.

Training the DL/ML model

The script file is the entrance of training different DL/ML models.