With the further spread of Japanese literature, this paper explores the application of artificial intelligence technology in Japanese literature education, improves the feature extraction TF-IDF algorithm based on information theory, counts the information content of the keywords as the control coefficient of their TF-IDF values in the final result, and constructs the TFIDF-BOIT linguistic feature extraction algorithm for Japanese literary works. The style analysis of literary works was conducted by screening five-dimensional variables based on the extracted linguistic features. A more targeted teaching path is proposed by analyzing the language features and styles of various Japanese literary works. The DR in Japanese literature during the Nara period was 0.345, and the DLS was 0.472. While in the Meiji period, the DR in literature decreased by 0.104, and the DLS increased by 0.085. The refinement of literature education can be achieved through the intelligent analysis of Japanese literature.