This paper establishes a BERT-BiLSTM-ATT-CRF named entity recognition model containing four levels, mines the semantic relevance of Chinese cultural entity-relationships, and constructs a knowledge graph ontology covering a large number of Chinese cultural resources. A collaborative filtering recommendation algorithm integrating students’ long- and short-term interest values and interest change degrees is proposed to improve the relevance and diversity of recommended resources based on the knowledge graph with students’ interests, and accurately support Chinese culture dissemination in the classroom teaching of overseas Chinese language teachers. The study shows that the knowledge graph constructed based on an ontology with more than 60% coverage of Chinese culture contains 7 major categories of cultural knowledge resources. Students’ short-term interest values in these 7 major categories of cultural resources range from 62.457 to 90.663, which are higher than the long-term values of 61.559 to 89.404, and the rate of interest change is faster. The recommendation algorithm that integrates students’ long-term and short-term interests is able to provide students with a recommended list of Chinese culture with an accuracy of 94.165% and a diversity of 88.132% at the same time. Using this algorithm, overseas Chinese language teachers are able to integrate Chinese culture knowledge resources in a timely manner to complete classroom teaching on the basis of grasping students’ current and long-term interests.
