This paper proposes knowledge representation based on knowledge graph embedding (TransE model) and based on deep wandering (DeepWalk model) to enhance the level of intelligent recommendation of knowledge points. Synthesize and construct a knowledge graph-based Civic Education model. Analyze the node centrality specifics of the model. Carry out a controlled experiment of model application and investigate student satisfaction on this basis. The three nodes with the highest node centrality are “life view and values”, “morality and law” and “patriotism and nationalism”. The average score of the test questions in the experimental class is 71.25, and the correct rate of the six types of test questions is higher than that of the control class. Most of the students’ satisfaction level with the intelligent teaching mode combined with the model was between 65 and 100 points. 92% of the students found the teaching mode interesting at a level between (75,100]. 90% of the students’ content mastery satisfaction level was between 85 and 100 points. Intelligent teaching using the knowledge graph-based Civics education model can help students improve their interest in learning Civics knowledge and construct Civics knowledge system.