Start Date
11-8-2016
Description
Existing researches have always focused on numerical data or textual contents in financial statements to expose a firm’s fraudulent behavior. To the best of our knowledge, this study has the first try to leverage huge amount of user generated contents in social media for fraud detection. A text analytic framework is created to decompose unstructured social media contents into words weights features, topic features and emotion features. With these social media features, social media contents prior to fraud disclosure of fraudulent firms and matched non-fraudulent firms can be classified with average accuracy at 81.43% under ten-fold cross validation. It demonstrates that there is a leading effect of social media contents for financial fraud disclosure. In addition, combining social media features with a set of financial ratios makes the fraud prediction accuracy at 83.57% on average. Investors, auditors, and policy makers can benefit a lot from the findings of this study.
Recommended Citation
Dong, Wei; Liao, Shaoyi; Xu, Yujing; and Feng, Xiaoqian, "Leading Effect of Social Media for Financial Fraud Disclosure: A Text Mining Based Analytics" (2016). AMCIS 2016 Proceedings. 2.
https://aisel.aisnet.org/amcis2016/AccountingIS/Presentations/2
Leading Effect of Social Media for Financial Fraud Disclosure: A Text Mining Based Analytics
Existing researches have always focused on numerical data or textual contents in financial statements to expose a firm’s fraudulent behavior. To the best of our knowledge, this study has the first try to leverage huge amount of user generated contents in social media for fraud detection. A text analytic framework is created to decompose unstructured social media contents into words weights features, topic features and emotion features. With these social media features, social media contents prior to fraud disclosure of fraudulent firms and matched non-fraudulent firms can be classified with average accuracy at 81.43% under ten-fold cross validation. It demonstrates that there is a leading effect of social media contents for financial fraud disclosure. In addition, combining social media features with a set of financial ratios makes the fraud prediction accuracy at 83.57% on average. Investors, auditors, and policy makers can benefit a lot from the findings of this study.