PhD defence Manal Laghmouch

Keywords: Audit, Information overload, Red flags, Machine learning

"Semi-Automatic Deviation Identification: The Way Forward to Assurance over Financial Information"

Auditors increasingly rely on process data to understand how business processes operate in practice and detect deviations from prescribed procedures. This creates new opportunities, but also new challenges: when every transaction is analyzed, the number of detected process deviations (i.e. red flags) becomes overwhelming. Only a small fraction of the detected red flags is truly relevant for the audit. Finding them is challenging and, in many cases, unmanageable. This dissertation starts from the premise that managing the overload of detected deviations is essential to unlocking the value of data-driven auditing. It introduces semiautomatic classification techniques that identify relevant deviations within large datasets. The findings point to a future in which human judgment and Machine Learning work together towards more effective audits.

Click here for the live stream.