As knowledge information becomes specialized, segmented, and complex, the construction of knowledge systems in the field of education management in colleges and universities becomes more and more difficult. The article extracts the knowledge in the field of education management through the entity relationship extraction model based on a convolutional neural network and then systematizes and summarizes the knowledge of education management by the knowledge network theory method to further build the framework of education management knowledge system. The centrality of the knowledge points in the education management system is calculated by the mediated centrality method and Dijkstra’s algorithm centrality calculation method, in which the centrality of the knowledge point k5 is the largest at 6.34. In the statistical analysis of high-frequency words in the knowledge system of education management, the vocabulary in the statistical table reaches the total number of 5,600 times, and the selected high-frequency words are able to reflect the main direction of the education management, which verifies the rationality of the knowledge system of education management proposed in this paper. The rationality of the proposed knowledge system for educational management.