Under the background of rapid development in the new media era, college students’ network clustering behaviors are becoming more and more significant, which has brought impacts on the healthy development of college students. Based on the analysis of the typical characteristics and emotional bias of college students’ network clustering behavior, the article establishes the basic framework of the evolution mechanism of college students’ network clustering behavior. With the support of structural equation modeling, an evolution model of college students’ network clustering behavior is designed to explore the factors affecting the formation of college students’ network clustering behavior. Using the “network stratification” event as a sample, we analyzed the evolution of university students’ network clustering behavior and proposed an optimization path. It is found that the model explains 73.45% of the college students’ network clustering behavior, and the network information elements have a greater influence on the college students’ network clustering behavior, among which the value identity has the most obvious influence (0.479). Therefore, it is necessary to start from the dimensions of controlling the initiative of network field, strengthening the education of network mentality guidance and improving the risk management mechanism of public opinion as a way to better regulate college students’ network clustering behavior.