Leveraging Advanced Machine Learning for Anomaly Detection in Graph Databases: a Focus on Fraud Detection in NoSQL Systems

EasyChair Preprint 13064

Abstract

Anomaly detection in graph databases stands as a critical frontier in fraud detection, particularly in domains such as social networks and financial transactions, where fraudulent activities manifest as subtle deviations from normal behavior. This abstract explores the application of advanced machine learning algorithms, specifically graph neural networks (GNNs) and community detection methods, within graph-based NoSQL databases to identify anomalous patterns indicative of fraudulent behavior.

Graph databases offer a natural representation of complex relationships and interactions inherent in fraud scenarios, making them well-suited for detecting anomalies. Graph neural networks, a class of deep learning models tailored for graph-structured data, excel at learning representations of nodes and edges while capturing intricate dependencies within the graph topology. By leveraging GNNs, organizations can detect anomalous patterns in graph-based NoSQL databases by identifying nodes or subgraphs exhibiting unusual behavior relative to their neighborhood.

Furthermore, community detection methods provide a powerful means of identifying densely connected subgraphs or communities within a graph. Anomalous nodes or edges that disrupt the cohesion of these communities may signify fraudulent activity.

Anomaly detection in graph databases stands as a critical frontier in fraud detection, particularly in domains such as social networks and financial transactions, where fraudulent activities manifest as subtle deviations from normal behavior. This abstract explores the application of advanced machine learning algorithms, specifically graph neural networks (GNNs) and community detection methods, within graph-based NoSQL databases to identify anomalous patterns indicative of fraudulent behavior.

Keyphrases: Graph Databases, Graph Neural Networks (GNNs), Machine Learning Algorithms, NoSQL systems, anomaly detection, community detection, financial transactions, fraud detection, graph topology, social networks

@booklet{EasyChair:13064,
  author    = {Dylan Stilinki and Kaledio Potter},
  title     = {Leveraging Advanced Machine Learning for Anomaly Detection in Graph Databases: a Focus on Fraud Detection in NoSQL Systems},
  howpublished = {EasyChair Preprint 13064},
  year      = {EasyChair, 2024}}