The article improves the traditional collaborative filtering algorithm, integrates it with the content-based recommendation algorithm, proposes the recommendation algorithm based on the mixture of collaborative filtering and content, and serves as the operation logic for designing the intelligent recommendation system of educational content for Civics class. Model variables are determined using structural equation modeling and relevant hypotheses are presented to construct a model of factors that influence student acceptance of the Civics Intelligent Recommendation System, followed by empirical analysis. The mean values of expectation performance, effort performance, social influence, convenience conditions, self-efficacy, perceived pleasantness, and willingness to use are 3.48, 2.70, 3.61, 2.36, 3.77, 3.84, and 3.73, respectively. Students’ use of the Civic Intelligent Recommendation System is greatly influenced by their perception of pleasantness and self-efficacy. The questionnaire has good reliability and validity in general. The initial model has valid hypotheses H1, H2, H6, H7, H8, H9, H10, and H11. In the analysis of variance, there were significant differences between genders only in performance expectations (0.000) and perceived pleasantness (0.016). Significant differences existed across grades in terms of performance expectations (0.018) and social influence (0.000). The measurement dimension of willingness to use had a moderating effect across majors. Hypotheses H12 and H13 are partially valid, but H14 is valid.