Research on Risk Classification of Online Public Opinion in Emergencies Based on FHCO Algorithm

Online public opinion concerning emergencies presents challenges to social stability due to its rapid dissemination and complex evolutionary nature. Accurate risk classification is essential for effective early warning. Current risk classification methods, however, face limitations such as difficulty in capturing complex structural features at a single scale, inadequate representation of local risk variations, and imprecision in threshold determination. To address these issues, this article proposes a risk classification method on the basis of the fractal hierarchical clustering optimization (FHCO) algorithm for public risk perception and discursive reflection. First, a comprehensive set of risk indicators for online public opinion is constructed. Next, fractal characteristics-including the correlation function, generalized dimension spectrum, singularity index, and fractal spectrum-are calculated. These fractal features are then clustered using the FHCO algorithm, and the final risk level is determined by integrating the comprehensive risk score. To validate the method's effectiveness, the "7.20" Zhengzhou rainstorm incident was analyzed as a case study, along with 27 other high-risk public opinion events. Results indicate that the proposed method achieves an accuracy of 92.86%, significantly outperforming the traditional K-means (64.29%) and K-medoids (82.14%) algorithms. This demonstrates that the method can effectively improve the accuracy of risk classification, providing new insights and technical support for the early identification of online public opinion risks associated with emergencies.