Assisting Telecommunication Fraud Prediction: Detect Individuals Carrying Multiple Phones Based on Trajectory Data Mining

Abstract

Telecommunication fraud grows rapidly in recent years, which brings serious property loss or even life loss to victims. However, investigation and evidence collection in such crime cases is extremely difficult, as telecommunication fraud committed by a group usually has the features of long-distance and non-contact. However, in known scenarios of telecommunications fraud, multiple mobile phones or SIM cards are essentially utilized for role pretending, identity hiding and increasing the success rate of scams. Meanwhile, these phones and cards may expose plenty of real-time location data to intentional investigators. Based on such observation, this paper gives a new way to detect telecommunications fraud by finding potential fraudsters based on trajectory data mining. Through analyzing trajectory data, individuals with multiple phones can be found and recognized as potential telecommunications fraudsters which should be intensively monitored. The trajectory clustering and FP-growth algorithm are adapted in the proposed method, and the effectiveness of the method is validated on real-world data sets and simulation data sets. The proposed method provides a technical support for the prevention of fraudulent activities.