Risk Assessment Based on Dataflow Dynamic Hypergraph for Cross-Border Data Transfer

In the era of big data, information security, particularly during data transmission, has become a top priority. However, existing risk assessments are often limited to local perspectives, missing a global view of data-related risks and hidden relationships. This paper introduces the Dataflow Dynamic Hypergraph (DDHG) framework, which abstracts data flow into a dynamic hypergraph, incorporating key features to better describe the data transfer process. To fully capture this information, we propose the Node-Edge-Hyperedge Autoencoder Graph Network (NEH-AGN), which frames risk assessment as a graph anomaly detection problem, identifying anomalous nodes for risk evaluation. The paper also explores future research directions in data flow scenarios, offering a reference for future work.