Project R-12261

Semi-automatically anomaly identification: the way forward to assurance over financial information (Research)

Financial data plays a vital role in economies as stakeholders base their decisions on the accuracy of financial information. To verify the accuracy of the financial statements, an independent auditor conducts an annual financial audit. To reach an opinion, the auditor has to check the financial transactions of the previous year. Since it is not feasible to manually check all of these transactions, the auditor relies on a sample. However, given current advances in information technology and data that are widely available, it is theoretically possible to leave the time perk of sampling and, instead, test all transactions. Unfortunately, automated testing of all transactions results in an overwhelming amount of alarms. Tens of thousands of alarms are presented to the auditor to investigate manually. The purpose of this project is to address the issue of too many alarms when conducting full-population testing and to develop a solution to investigate all raised alarms in an efficient way. The project will design three versions of a classification technique that automatically distinguishes 'real' alarms from acceptable exceptions in a business process context. As such, the project goes well beyond the current academic state-of-the-art and creates innovative new tools and insights for our current and future auditing.

01 November 2021 - 31 October 2023