The current social security situation remains generally stable. However, with the emergence of new technologies such as the Internet, there has been a continuous increase in trending incidents. In this context, it becomes imperative to systematically summarize various types of events and extract characteristic patterns of different cases, thereby providing targeted operational solutions for public security authorities when handling similar incidents. This study collects five years of case data from public security enforcement records and employs both K-Means clustering algorithm and LDA topic modeling to analyze and categorize all cases while identifying their thematic features. When new incidents emerge, a Naive Bayes classifier can be implemented to categorize the case into existing clusters, enabling law enforcement agencies to apply established response protocols from similar historical cases. This methodology facilitates data-driven decision-making through systematic pattern recognition and predictive classification.