Under the rapid development of digital education, the introduction of artificial intelligence technology into the reform of language education can make up for the shortcomings of traditional methods and improve the quality of language education. In this paper, we first design the joint extraction algorithm of entity relationship based on graph convolutional neural network, which improves the accuracy of recognition by fusing the designed ontology attributes. Then it proposes the method of extracting the features of language education entities in the knowledge graph using R-GCN model for classifying and predicting the potential relationships between entities, so as to realize the construction of the knowledge graph of language subjects. Finally, a reform path for language education is proposed, with a view to providing a useful reference for promoting language education to a higher level. In the dissimilarity matrix analysis of high-frequency keywords in language education research. It is found that the relationships with language education are, in descending order, language teaching, comprehensive learning, language education, curriculum standards, language teachers and so on. Thus, it can be concluded that “language education” and “language teaching” are most closely related in the research process of language education.
