Based on the feasibility of a personalized teaching platform and user requirements, this paper puts forward the overall architecture design and database design scheme of a personalized teaching platform in ideological and political education, which mainly consists of three functional modules, namely, the knowledge mapping module, ideological and political education course recommendation module, and learning effect evaluation module. After crawling the initial data based on the LTP model, the ideological and political education course resources were extracted and integrated to complete the construction of the knowledge graph module. The ideological and political education course recommendation module is created using the KGCNN algorithm, and then the learning effect evaluation module is constructed by combining the online behavior of students. After testing the system’s performance, the application effect of the teaching platform is assessed. The results show that KGCNN aggregation layer space in the interval of 10~100 can embed data with power law distribution more effectively, and the KGCNN algorithm also has certain advantages in the field of modeling personalized teaching platforms for ideological and political education. The number of experimental classes and ordinary classes with final grades in the L1 band increased by 21.90% and 7.17%, respectively, compared to the midterm, indicating that the personalized teaching platform for ideological and political education courses can effectively promote the improvement of student’s academic performance, and the enhancement effect of students with high levels is more significant.