부산대학교 도서관

In recent years, with the major transformation of the socio-economic landscape, various types of conflicts and disputes have become more frequent and widespread. A significant number of these conflict and dispute-related complaints have resulted in unresolved cases and difficult-to-handle situations due to unclear responsibilities and insufficient services. The complexity and diversity of conflict and dispute issues pose challenges to the early warning system. Different types of conflict and dispute events may have distinct characteristics and patterns, requiring diverse algorithms and models for identification and prediction. This article proposes an intelligent early warning system based on the integration of improved GRU (Gated Recurrent Unit) and CNN (Convolutional Neural Network). By combining CNN and GRU, this system addresses the issue of predicting conflict and dispute cases, actively resolving conflicts, and preventing the escalation of conflict and dispute risks to a red alert level. Through a comprehensive risk analysis model, it can effectively identify high-risk conflict and dispute cases, facilitating relevant departments to take timely preventive and management measures, thereby maintaining social stability.