ABSTRACT
ABSTRACT: Early warning of financial risk is an important research direction in financial data mining. Financial risk warning is difficult because financial data are various, complex, and dynamic. Using data mining technology can discover the hidden abnormal transaction information from massive financial data, and then the technology can monitor and deal with it a timely manner. Early warning of financial risk can reduce the operational risk of financial institutions effectively. This paper introduces a method of discovering suspicious transaction information at risk with early warning. The method is based on the non-instructional link discovery technology in data mining. It can realize the hidden information found in massive data. The method first constructs the financial transaction network. Second, it finds the target node and path of the transaction, and then filters them. Then it calculates the transaction frequency of the selected path. Finally, it finds the outlier based on the distance-based outlier detection algorithm. These outliers are the suspicious transaction information. By processing these information, financial institutions can avoid financial risks.
