This research focuses on the problem of rumor detection in social media platforms. With the rapid spread of information on the network

rumors can spread rapidly through the interaction between users

which leads to the spread of misinformation and has a negative impact on society. Therefore

how to effectively identify rumors has become an important research topic.This paper proposes a modeling method based on graph structure to describe the information dissemination process. Specifically

social media data is represented as a graph structure

in which nodes represent users or posts

while forwarding

commenting and other interactive relationships. On this basis

the Graph Convolutional Network (GCN) is used to learn the representation of nodes. By aggregating the information of neighboring nodes

the model can use text features and disseminate structural information at the same time.The evaluation of the proposed method through experiments shows that compared with the method that relies only on text information

the introduction of communication structure can effectively improve the performance of rumor detection.Overall

this study provides a structured method for rumor communication modeling and reflects the application value of graph-based models in social media analysis.