With the continuous development of innovative education and the growth of the demand for personalized learning, this paper explores the evaluation of students’ learning ability and resource recommendation methods in Civic Education under the guidance of OBE (student-centered, outcome-oriented, continuous improvement) educational philosophy to better achieve the educational goals. By constructing a learning ability evaluation model for Civic Education, this paper establishes a learning ability evaluation and Civic Education resource recommendation model based on Bayesian decision making by using Bayesian Decision Theory’s a posteriori probability for uncertainty reasoning and combining the personalized recommendation function of collaborative filtering algorithm. By analyzing the data of a university’s ideological and political education course in the fall of 2020, this paper not only realizes the accurate evaluation of students’ learning ability, but also tests the model’s effectiveness in learning ability evaluation and resource recommendation. The findings show that students perform well in effective communication, collaboration and independent learning ability, whose evaluation mean values are 0.62, 0.58 and 0.55, respectively. In addition, some students’ learning ability significantly improves after receiving the recommended test questions, with the increase ranging from 34.36% to 36.67%. These findings demonstrate the model’s excellent learning ability evaluation and resource recommendation performance. The research in this paper provides valuable references for evaluating students’ learning ability and resource recommendation in Civics and Political Science education, which can help further improve students’ learning effectiveness and achieve educational goals.