RIFT 2026 Hackathon - Graph Theory Track
Money muling is a key technique used in financial fraud, where illicit funds are transferred through multiple accounts to obscure their origin. Traditional database queries fail to detect these complex, multi-hop transaction patterns.
This system models financial transactions as a graph and detects:
- Circular fund flows (fraud rings)
- High-centrality accounts (potential hubs)
- Suspicious transaction patterns
Instead of analyzing transactions in isolation, this project uses graph theory to analyze relationships between accounts.
- Uses Tarjan’s SCC algorithm for fraud ring detection
- Detects multi-hop laundering patterns
- Designed for large-scale transaction datasets
- Backend: Python (FastAPI)
- Graph Engine: NetworkX
- Database: MySQL
- Visualization: Cytoscape.js
Input: Transaction dataset (CSV)
Output:
- Flagged accounts with suspicion scores
- Identified fraud rings
- Visualization of transaction graph
- Circular Routing (90-100%): Accounts found in a closed directed cycle (length 3-5) are flagged with the highest risk.
- High Centrality (70-85%): Accounts acting as "hubs" (high degree centrality) are flagged for Smurfing/Layering patterns.
- Pattern Weights: Scores are calculated based on the account's position within the network and the density of the identified fraud ring.
-
Cycle Detection: Tarjan’s SCC algorithm -
$O(V + E)$ -
Centrality Analysis: Degree Centrality -
$O(E)$ - Execution: Optimized for sub-second processing of large transaction CSVs.
- Clone the repository.
- Install dependencies:
pip install -r requirements.txt - Configure MySQL in
database.py. - Run the app:
python main.py - Access the dashboard at
http://127.0.0.1:8000
- Anirudh Dhamodaran
- Jithesh Sankarganesh
- Chris Johnson
- Darshan E