Research and Development of Business Intelligence Technology for Adaptive Fraud Management System

Description

Fraud management is an important R&D topic in telecom industry. According to a survey of Communications Fraud Control Association (CFCA http://www.cfca.org), the annual worldwide telecom fraud losses were US$12 billion in 1999 and they rose to US$40 billion in 2003. As the next generation network (NGN, e.g. WCDMA, CDMA 2000 and TD-SCDMA) is launched to offer new content services, fraud problems become a major concern to mobile and wireless telecom operators today. Currently, most of the fraud management systems (FMS) are developed using rule-based reasoning, balanced score card and neural network technologies, however they can hardly meet the growing demands for new content services on NGN. This project aims to develop new business intelligence (BI) technology for implementation of an adaptive FMS. It combines advanced artificial neural networks, fuzzy logic reasoning, genetic algorithms, automated rule extraction and multi agent technologies to achieve a new level of decision intelligence. The resultant FMS will have the most advanced capabilities for early detection of new frauds, fast and effective fraud prediction and high quality referrals to case investigation. It is to detect the frauds at an early stage so as to reduce the potential revenue losses. The project results can be used by mobile and wireless telecom operators and system integrators