The increasing development and maturity of Artificial Intelligence Generated Content (AIGC) provides a solid technical support for the application of image narrative model in Jilin Party History. This paper constructs an image generation model with StyleGAN2 image generation model and ChineseCLIP model as the main components. The model is divided into a training phase and a testing phase, in which the ChineseCLIP image encoder is used to extract image features, and the ChineseCLIP text encoder is used to extract the input Chinese text features in the testing phase, thus realizing an accurate understanding of the input Chinese semantics. The spectral normalization technique is added to the model discriminator to constrain the model gradient value, which stabilizes the model training while maintaining the original expression ability of the model. In addition, the social network analysis method is adopted to carry out the quantitative analysis of the content of Jilin Party history topics by selecting the overall structure of the network, the characteristics of subgroups, key nodes and other corresponding indicators. Based on the image generation model of this paper, the image narrative communication network structure of Jilin Party history, with a graph density of 1.42, modularity coefficient of 0.931, and an average path length of 1.81, not only has a significant network community structure and a high efficiency of information dissemination, but also is an effective path for AIGC technology to assist in the innovation and development of the spirit of party history in Jilin, and to promote the inheritance of the spirit of party history.
