Latent space network model for the popularity effect, with applications to Bitcoin networks.

We propose a dynamic latent space network model that incorporates the popularity effect, offering a unified framework to simultaneously account for the popularity and proximity effects in network dynamics. The model represents node attributes as latent positions, enabling the analysis of connectivity patterns influenced by both proximity and popularity while quantifying their relative contributions. A Bayesian inference algorithm is developed to estimate model parameters and latent positions, and its effectiveness is validated through simulation studies. The proposed model is applied to Bitcoin trust networks (OTC and Alpha), revealing key insights into their structural evolution and the distinct roles of popularity and proximity effects. Our findings demonstrate the versatility of the model in capturing temporal dynamics and supporting applications such as node recommendation and network visualization.