/GNN-Call-Graphs

Application of Graph Neural Networks on Call Graphs of Software

Primary LanguageJupyter Notebook

A call graph is a control-flow graph, which represents calling relationships between subroutines in a computer program. Here we have tried to predict which function or subroutine would call other function or subroutine using Graph Neural Networks. Following GNN models are used:

  • Graph Convolutional Network
  • GraphSage
  • Graph Isomorphism Network

Different features used for the task are:

  • Centrality measures as node information
  • Function name, parent class, parent package and structural information
  • Temporal information i.e. the sequence in which function call each other

To create enviornment run

conda install mamba -c conda-forge -y
mamba init bash
mamba env create -f environment.yaml

To activate env

conda activate callgraphs

To see the updated results and recreating the model in your local system

  • Setup the Anaconda environment
  • Run the individual scripts for the 3 GCN Models