In this paper, a relatively complete digital education management information system is established for the digital basic education resources involved in the field of basic education of Marxist theory to realize the digital transformation of party history and party-building education. Faced with a large amount of student data, the improved new BP model is utilized for integration, cataloging, and prediction. The data elements describing the resources are used as semantic tags to generate XLM documents in a standardized format, which enables the system to exchange data with other resource systems easily. It can be observed that the BP model accurately predicts and analyzes the shared learning behaviors of certain students with a prediction accuracy of 98% upon analysis. The educational management information system proposed in this paper can effectively identify potential risk-takers and categorize learning groups. The indicator of resource-sharing learning behavior has a significant effect on the learning effect of Marxist theory (P<0.05).