This paper discusses the ways in which big data analytics and behavioral biometrics can prevent or detect digital fraud. We used Random Forests and Neural Networks as our machine learning models to learn the dynamics of users' keystroke, mouse movement, and touch interaction patterns to identify fraud. It enhanced the relevancy and utility of the data and made the database algorithms and queries easier to manage over half a terabyte of online source raw data by applying data pre-processing and normalization to reduce noise, standardize data format, and make the data more relevant. This paper reveals that a behavioral observation system and a concept of big data that can be followed in real-time greatly enhance the fraud detection methodology and system flexibility. Another contribution of the paper is that such continuous user authentication will reduce the level of intrusiveness and improve security. Related literature indicates that our technology is comparable to practices in other fields and provides a versatile way of fighting new and advanced fraud. The idea of increased collaboration between industries and improvements in AI algorithms to detect fraud replaced the potential data privacy, compliance and computational power limitations. The study validates the notion that additional actionable plans are needed to combat next generation fraud and improve online security.
Abbas et al. (Sat,) studied this question.