AutoAudit: Mining Accounting and Time-Evolving Graphs

---

Lee, MC., Zhao, Y., Wang, A., Liang, P.J., Akoglu, L., Tseng, V.S., and Faloutsos, C., AutoAudit: Mining Accounting and Time-Evolving Graphs. IEEE International Conference on Big Data (Big Data), 2020.

https://ieeexplore.ieee.org/abstract/document/9378346

Please cite the paper as:

@inproceedings{lee2020AutoAudit,
  title={{AutoAudit:} Mining Accounting and Time-Evolving Graphs},
  author={Lee, Meng-Chieh and Zhao, Yue and Wang, Aluna and Liang, Pierre Jinghong and Akoglu, Leman and Tseng, Vincent S. and Faloutsos, Christos},
  booktitle={2020 IEEE International Conference on Big Data (Big Data)},
  year={2020},
  organization={IEEE},
}

Introduction

In this paper we propose AutoAudit, a systematic method for handling anomaly detection problems not only in accounting datasets, but also in other real-world datasets. It consists four major components:

• "Smurfing" Detection: We proposeAA-SMURF, an un-supervised and parameter-free algorithm to detect injected“Smurfing” pattern in real-world datasets.
• Attention Routing: We proposeAA-ARto attend to themost suspicious periods in time-evolving graphs and pro-vide explanations.
• Discoveries: We discover three month-pairs with highcorrelation, proved by “success stories”, and patterns ofaccounting datasets follow Power Laws in log-log scales.
• Generality: We further generalized our method on otherreal-world graph datasets, such as Enron Email and CzechFinancial datasets.

Installation and Dependency

The experiment code is writen in Python 3 and built on a number of Python packages:

• matplotlib==2.0.2
• pandas==0.21.0
• scipy==0.19.1
• numpy==1.13.1
• scikit_learn==0.19.1

---

Datasets

Three datasets are used (see dataset folder):

Datasets	Nodes	Edges	Time Span
Accounting	254	285,298	01/01/2016 to 02/06/2017
Czech Financial	11,374	273,508	01/05/1993 to 12/14/1998
Enron Email	16,771	1,487,863	01/01/2001 to 12/31/2001

• Czech Financial dataset can be found in https://data.world/lpetrocelli/czech-financial-dataset-real-anonymized-transactions.
• Enron Email dataset can be found in https://www.cs.cmu.edu/~./enron/.

Usage and Sample Output

Experiments could be reproduced by running the code directly. You could simply download/clone the entire repository and execute the code by

python AA-Smurf.py
python AA-AR.py

Conclusions

In this work, we present AutoAudit, which addresses the anomaly detection problem on time-evolving accounting datasets. This kind of data is usually complicated and hard to organize. Our main purpose is to automatically spot anomalies, such as money laundering, providing huge convenience for auditors and risk management professionals. Our approach is also general enough to be easily modified to solve problems in different domains.