Evaluating the Feasibility of Using Classifiers in Detecting Social Engineering Fraud

Abstract

Social engineering fraud is among the most notorious forms of fraud through which people continue to lose money and sensitive information. Its increasing prevalence is negatively affecting strides made in mobile and digital banking. Despite efforts in creating public awareness, its mitigation has not been effective as the tricks used by swindlers keep evolving. Virtually all existing solutions to the problem are based on human interventions such as manually reporting and blacklisting phone numbers. This approach is slow and inefficient due to the huge number of incidents reported relative to the limited existing human resource capacity. This paper presents an evaluation of the feasibility of using classifiers to detect voice-based social engineering fraud. Findings suggest the possibility of using natural language processing and machine learning to automate the detection of voice-based social engineering fraud. Outcomes of the study can be used to develop automated real-time SEF detection systems.