UGFraud
by safe-graph
UGFraud is an unsupervised graph-based toolbox that detects fraud by analyzing suspiciousness in nodes and edges within bipartite graphs using state-of-the-art algorithms.
An Unsupervised Graph-based Toolbox for Fraud Detection
Primary Use Case
UGFraud is designed for data scientists and security analysts who need to detect fraudulent activities in graph-structured data such as user-product interactions without labeled data. It is particularly useful for identifying anomalies and assessing risk in networks where relationships between entities can reveal suspicious behavior.
- Unsupervised fraud detection using graph-based algorithms
- Supports bipartite graphs like user-product graphs
- Estimates suspiciousness scores for both nodes and edges
- Implements Markov Random Field (MRF)-based, dense-block detection, and SVD-based algorithms
- Minimal input requirements: graph structure and optional prior suspicious scores
- Integration with other graph-based fraud detection tools like DGFraud
- Open for contributions and extensible with new fraud detectors
- Provides citation and academic references for research use
Installation
- pip install UGFraud
- git clone https://github.com/safe-graph/UGFraud.git
- cd UGFraud
- python setup.py install
Usage
>_ pip install UGFraudInstalls the UGFraud toolbox from PyPI.
>_ git clone https://github.com/safe-graph/UGFraud.gitClones the UGFraud repository from GitHub.
>_ cd UGFraudChanges directory to the cloned UGFraud folder.
>_ python setup.py installInstalls UGFraud from the cloned source.
- Integrate UGFraud with SIEM platforms to enhance anomaly detection in graph-structured data.
- Use UGFraud outputs to prioritize alerts for security analysts, reducing false positives in threat hunting.
- Combine UGFraud with graph neural network tools like DGFraud for layered fraud detection strategies.
- Leverage UGFraud in purple team exercises to simulate and detect complex fraud attack patterns.
- Automate suspiciousness scoring to trigger adaptive risk-based access controls in real-time.
Docs Take 2 Hours. AI Takes 10 Seconds.
Ask anything about UGFraud. Installation? Config? Troubleshooting? Get answers trained on real docs and GitHub issues—not generic ChatGPT fluff.
This tool hasn't been indexed yet. Request indexing to enable AI chat.
Admin will review your request within 24 hours
Related Tools
DGFraud
safe-graph/DGFraud
A Deep Graph-based Toolbox for Fraud Detection
CreditCardFraudAnalysis
Kanav-code/CreditCardFraudAnalysis
Prediction of the credit card payments done by clients using various Classification models.
fraud-report
juancarbajal/fraud-report
Sistema de detección de fraudes en una base de datos de compras.