Currently, the teaching of Civics in engineering management courses faces the problems of difficulty in selecting teaching materials and lagging behind in content updating, and teachers have invested a lot of time and energy in integrating professional knowledge with Civics elements. The rapid development of natural language processing technology provides technical support for solving these problems and realizes the efficient automated generation of Civics teaching content for engineering management courses by automatically filtering and matching relevant Civics elements through algorithms. In this study, we first constructed the database and knowledge graph of Civics in engineering management courses, and designed the automatic screening algorithm for Civics elements. Then the keywords are extracted using the TF-IDF algorithm, and the BERT and GPT-2 models are used to generate the Civics text content. Finally, the generation effect is evaluated by content quality score, keyword coverage and student score improvement rate. The results show that the percentage of content with 0.4 to 1.0 quality scores reaches 74.4%, and the students’ scores are improved by 7.2% compared with the traditional textbook after teaching Civics content based on natural language processing technology. The Jaccard similarity coefficient test shows that the average value of corpus overlap rate under the same topic is 28.71%, and the overlap between different topics is 29.31%. The results of the student feedback experiment show that the difficulty coefficient of the test paper is 0.64 falling in the moderate category, and the total average score of the content difficulty is 4.175.This study proves that natural language generation technology has a good prospect of application in the automated generation of the content of the teaching of Civics and Politics in engineering management courses, and it can effectively improve the quality of teaching and students’ learning effect.
