Abstract This study analyzes SCIE papers published by 39 universities as a substantial big-data sample, employing latent class analysis to explore the characteristics of university-enterprise cooperation in Chinese institutions. It identifies five categorical attributes to construct a latent class model. This paper integrates hierarchical analysis with fuzzy mathematics to develop a comprehensive fuzzy evaluation model. By organizing and layering the factors, a rational set of indicators is established, and a hierarchical analysis structure is constructed. Utilizing a judgment matrix, the study calculates the importance rankings of all indicators, thereby highlighting the significant influence of the factors. Empirical results indicate that the model achieves optimal fit with five latent categories. Within the five primary indicators for evaluating the innovation capability of university-enterprise cooperation, the weight of innovation outcomes is the highest at 0.24, followed by industrial innovation at 0.22. The analysis reveals a regional imbalance in university-enterprise collaboration across the eastern coastal area, with Guangdong achieving the highest rating of 84.03, closely followed by Jiangsu. In assessing the innovation capability of university-enterprise cooperation in Guangdong's applied universities, the overall scores for universities categorized as B, C, and D are 3.641, 3.522, and 3.409, respectively. Notably, the scores for R&D investment intensity and Chinese patent applications are relatively low, suggesting these areas require enhancement to bolster support and development, thereby improving university-enterprise cooperation and innovation capabilities.