关键词: 双模网络, 潜在空间模型, 流行度, 链路预测

Abstract: In order to study the probabilistic characteristics of bipartite network, we propose a popularity-scaled latent space model for bipartite network. It considers various network features, such as heterogeneity and transitivity. Furthermore, these features can be expressed and explained explicitly under the model assumptions. Based on the model, a weighted probability index is proposed to measure the possibility of future connection between nodes in the bipartite network. To demonstrate the finite sample performance of the index, abundant simulation studies are conducted. In addition, both a public dataset and a real dataset are analyzed. Experimental results show that the proposed method can not only quantify the characteristics of the network generation process, but also outperform other link prediction methods for bipartite network.

Key words: Bipartite Network, Latent Space Model, Popularity, Link Predication