Uncovering Hidden Social Connections in Lawsuit Activities by Learning Over Knowledge Graph

While frequent lawsuits against businesses for claiming excessive damages have negative impacts on the economy, lawsuits may be strategically employed to improve law enforcement and social governance. Chinese consumer citizen suits are an example of such, although their patterns and impacts remain a mystery. This work constructs a comprehensive knowledge graph (KG) from judicial decisions and employs transformer-based heterogeneous graph neural networks (HGNNs) to analyze the behavioral patterns of plaintiffs. By leveraging deep contextual features from the KG, this study uncovers hidden social relationships among plaintiffs, revealing organized litigation behaviors facilitated by shared legal resources, such as lawyers, law offices, and common defendants. To the best of our knowledge, this is the first empirical study to uncover and validate hidden social connections among consumer-plaintiffs in China using a KG and a transformer-based HGNN framework. The findings reveal significant community structures, indicating that plaintiffs may frequently act in organized groups in lawsuit activities. This study advances methodological approaches by integrating HGNNs for link prediction and modularity-based community detection, offering actionable insights into the dynamics of grassroots litigation. By introducing a novel analytical framework and dataset, this work deepens the understanding of consumer lawsuits, underscores the influence of social networks on litigation behavior, and lays a foundation for future research in legal analytics and social network analysis (SNA) within the legal domain.