This paper focuses on the method of content recommendation and influence analysis based on matrix decomposition in cross-cultural social network communication. Through in-depth study of the basic parameters of social network structure and methods such as matrix decomposition, it provides a solid theoretical foundation for the construction of influence calculation and content recommendation model. User rating information and social network structure information are combined to calculate the influence between users, and the influence between users and users’ personal influence are combined to calculate the influence between asymmetric users. On the basis of the SoReg method, this paper proposes the SoInf recommendation model based on matrix segmentation and integrating community structure and social influence, so as to utilize the influence information between users and user rating information for recommendation. The results show that the index values of the influence calculation method in this paper show good correlation with the real spreading influence of individual nodes, and the Top-5 nodes calculated by this paper’s method have stronger spreading ability compared with the nodes derived by other methods. In addition, the SoInf recommendation model constructed in this paper has higher accuracy and recall as well as lower Mean Absolute Error (MAE) than the benchmark model that does not consider the social network structure.
