Abstract
Telephone scams exact a huge toll on consumers and society, with at large annual costs in the Taiwan alone can exceeding $4 billion. The global proliferation of the Internet has enabled con artists to export their craft to a rapidly expanding market and reach previously untapped consumers. For example, fraud gangs can make a call from China using the fake phone number to Taiwan. People in Taiwan will average received a fraud phone each month. In this paper, we analyze the call detail records and Blocked record of the mobile phone user. More precisely, we identify the fraud calls and analyze the calling behavior of these scammers. To detect scammers, we use statistics and social network analysis, developed four features: Call Count, Call Friend, Call Relevance and Call Duration Interval. Based on the analysis of calling behavior, we build an algorithm. Our results show that this model can detect and predict abnormal calling behavior with high precision (76.1%). We can block fraud phone numbers faster than ever before.