Matrix Representation of Graph Probability Diffusion Model Based on Heterogeneous Social Network

As a current research focus and hotspot, the graph probability diffusion model has been widely used in recent years because of its solid physical meaning and excellent recommendation performance. However, how to unify and represent the diffusion process of the model from the view of matrix theory? This issue is of great significance for unifying the algorithm representation and facilitating the model calculation. To address this challenge, this paper discusses the graph probability diffusion model from the perspective of matrix representation. Specifically, we first elaborate the purpose and purport of the graph probability diffusion model. Then we describe the matrix representation of the diffusion and recommendation process of this model. Finally, we give a concrete application example to validate the feasibility of the proposed method.