mhaseebtariq/fastman

FaSTMⱯN

Follow all Suspicious Trails of Money for all Nodes

The implementation is tailored for running things in a distributed and scalable way. With minimal changes the code can therefore, be executed on a massively large dataset.

The attached notebook fastman.ipynb contains the example dataset and the implementation for the following parts of FaSTM∀N:

• Temporal Graph Creation
• 2nd Order Graph Creation
• Weights Calculation
• Community Detection

! pip install matplotlib==3.4.3
! pip install networkx==2.6.3
! pip install pandas==1.4.1
! pip install pyspark==3.2.0
! pip install graphframes==0.6
! pip install leidenalg==0.9.1

The Example Dataset and the Resulting Graphs

Temporal Graph of Sequential Transactions

Second Order Graph

NOTE: The weights will be calculated per distinct moment

This means, for instance, [t1]A>>B ---> [t3]A>>B ---> [t4]B>>C is counted once

Detected Communities