INTERNET FINANCIAL FRAUD DETECTION BASED ON A DISTRIBUTED BIG DATA APPROACH WITH NODE 2 VECTOR

Abstract

The rapid advancement of information technologies, including the Internet of Things (IoT), Big Data, Artificial Intelligence (AI), and Blockchain, has significantly transformed consumer behaviors and reshaped the financial industry's development model. While the integration of financial services with these new technologies has offered consumers unparalleled convenience and efficiency, it has also introduced new and concealed risks of fraud. Instances of fraud, arbitrage, and predatory practices have had detrimental effects and led to substantial losses within the realm of Internet and IoT-based finance. Moreover, the exponential growth in financial data poses challenges for traditional rule-based expert systems and conventional machine learning models, making it increasingly arduous to detect fraud within vast historical datasets. Additionally, the rising specialization of financial fraud schemes enables perpetrators to evade detection by constantly evolving their tactics. To address these challenges, this article proposes an intelligent and distributed approach leveraging Big Data techniques for detecting Internet financial fraud. Specifically, the method employs the graph embedding algorithm Node2Vec to capture and represent topological features within financial network graphs as low-dimensional vectors. These representations facilitate intelligent classification and prediction of data samples using deep neural networks. The approach is implemented in a distributed manner on Apache Spark GraphX and Hadoop clusters to enable parallel processing of large datasets. Experimental results demonstrate that this approach significantly enhances the efficiency of Internet financial fraud detection, yielding improved precision, recall rates, F1-Score, and F2-Score metrics.