With the rapid development of mobile Internet technology and social media software technology, it has an important impact on the dissemination of Japanese text information. In this paper, Japanese text is conventionally preprocessed, and the LDA topic model is selected to represent Japanese text, and then the feature extraction of Japanese text is accomplished with the help of the chi-square distribution, and the features are deployed into the plain Bayesian algorithm for classification. Taking this as an entry point, the Japanese text-user interaction is established by introducing the similarity degree, and the information diffusion model is formed with the support of conditional random field theory and feature function. Compared with the traditional method, the prediction effect of the Japanese text information dissemination path of the model in this paper is particularly outstanding, with the values of 0.7439, 0.7743, and 0.7758, which verifies the performance of the information diffusion model based on the conditional random field in the application of the information dissemination path of the Japanese text, and it has an optimizing effect on the current information dissemination path of the Japanese text.
