Unveiling Deception : Cutting-Edge Approaches for Fraud Detection in Financial Statements

A wide range of parties involved in evaluating economic performance, including investors, auditors, interest groups, bankers, government organizations, academics, and analysts, can benefit greatly from the immediate detection of fraudulent financial reports. This research presents a novel methodology for the rapid detection of fraud in Greek companies. To detect potential fraud in Greek financial reports, we use computational natural language processing and machine learning.Our analysis is based on a dataset containing English-language annual reports of Greek companies from 2002 to 2015. This dataset includes 525 annual reports of Greek companies traded on the Athens Stock Exchange, representing a variety of industries, apart from the banking, utilities, and financial services sectors.Using three different text encoding approaches, we evaluated five computational intelligence models to search for likely fraud in these annual reports. The results of our experiments show that the proposed method has a hit rate of up to 93% in detecting fraud in the annual reports of Greek companies