Hello, and thank you for releasing this work.
In the MalGraph paper, the CFG are extracted using IDA Pro and I would like to ask a clarification question.
Is it feasible to extract the same type of:
- control-flow graph (CFG)
- call graph
- function boundaries
using Ghidra, instead of IDA Pro?
- Does your pipeline depend on any IDA-specific metadata or heuristics that would make Ghidra incompatible?
- Have you (or others) tested Ghidra-generated CFGs with your model?
- Would replacing IDA with Ghidra primarily affect graph quality, feature consistency, or model robustness?