Accounting irregularities raise significant risk for corporations and investors. Most of the well-known techniques for detecting accounting irregularities operate at the financial statement level. We show how exploiting the link structure be-tween accounts has the potential to greatly increase the accuracy of classification methods. We introduce SNARE(Social Network Analysis for Risk Evaluation), which detects related entities that may be overlooked by using in-dividual risk scores.
Auto117
https://www.cs.cmu.edu/afs/cs.cmu.edu/Web/People/mmcgloho/pubs/snare.pdf