Social network false information propagation tracing method based on graph neural network

Abstract

The invention discloses a social network false information spreading traceability method based on a graph neural network, which comprises the following steps: S1, constructing a multi-level situation graph model, expressing user nodes in a social network and a connection relationship thereof as a multi-level situation graph structure, and constructing each situation layer according to situation factors such as time, place and user psychological state; s2, designing and training a multi-level context awareness dynamic graph neural network, performing information fusion and interaction among different context layers, learning node embedding representation and adjusting a traceability path; s3, introducing a cooperative confrontation model driven by a game theory, and generating a preliminary false information propagation path; s4, optimizing a path by using a swarm intelligence algorithm and positioning a false information source node; s5, generating and marking a traceability path diagram of the key node and the source node; s6, when the social network dynamically changes, updating the model and the path; and S7, outputting the optimized traceability path diagram. According to the method, the accuracy and the real-time performance of false information tracing in the complex social network are improved.

Images