Occupational Fraud Detection Through Visualization
This addresses the challenge of internal fraud detection for companies, but it is incremental as it builds on existing visualization techniques for specific fraud scenarios.
The paper tackles the problem of detecting occupational fraud by processing large amounts of textual data from company systems, presenting a visualization-based system that uses a spiral layout to highlight suspicious events based on time-stamps and entity rankings, resulting in automated support for internal auditors.
Occupational fraud affects many companies worldwide causing them economic loss and liability issues towards their customers and other involved entities. Detecting internal fraud in a company requires significant effort and, unfortunately cannot be entirely prevented. The internal auditors have to process a huge amount of data produced by diverse systems, which are in most cases in textual form, with little automated support. In this paper, we exploit the advantages of information visualization and present a system that aims to detect occupational fraud in systems which involve a pair of entities (e.g., an employee and a client) and periodic activity. The main visualization is based on a spiral system on which the events are drawn appropriately according to their time-stamp. Suspicious events are considered those which appear along the same radius or on close radii of the spiral. Before producing the visualization, the system ranks both involved entities according to the specifications of the internal auditor and generates a video file of the activity such that events with strong evidence of fraud appear first in the video. The system is also equipped with several different visualizations and mechanisms in order to meet the requirements of an internal fraud detection system.