Research on Dispute Case Relationship Analysis and Prediction Based on Graph Convolutional Network (GCN)

In this paper, a method combining Graph Convolutional Network (GCN) and BERT model is proposed to analyze and predict the relationship between dispute cases in depth. By constructing a case relationship graph and applying GCN to learn the complex relationship patterns in the graph, and using BERT to capture the deep semantic information contained in the judgment documents, the comprehensive and deep learning of case features is realized, which innovatively combines the structural information processing ability of GCN and the text understanding advantages of BERT. A comprehensive model that can accurately predict the development trend and relationship of cases is constructed, which shows significant advantages especially in dealing with highly related and semantic complex dispute cases.