Financial Fraud Detection Using Social Media Cues ─ A Comprehensive Survey, Prediction, and Prototype Development

Abstract

Financial fraud has been a serious concern for investors, auditors, government regulators, and various stockholders that interact with public companies. Given that social media has become a popular venue for companies and management to broadcast product news and company-related updates, fraudulent financial information could be transmitted via social media platforms. Content generated on social media platforms expands the richness of the fraud feature sets with appropriate consideration of the context surrounding the company. However, existing financial fraud detection studies have typically utilized feature sets derived as non-holistic snapshot from firms’ financial statements, with little focus on structural and contextual features from social media networks. This research gap is attributable to the information complexities associated with social networking. To extract new and existing features for financial statement fraud detection, we propose a design framework for social media-based financial information analysis. We utilize the business network analysis to detect latent fraud features; a framework is designed to identify fraud features from various social media content. The experimental results confirm that the business network analysis could contribute to relation-based (i.e., collaborative and competitive) feature selection as pioneering fraud evidence for decision-making. The framework advocates the development of systems that are capable of providing rich information types inherent in social media networks. It also provides guidelines regarding the choice of features.

Date of Award	12 Nov 2015
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Shaoyi Stephen LIAO (Supervisor)