Optimization and application of multifeature fusion similarity indices in hyperlink prediction

Traditional link prediction methods primarily focus on the relationships between two nodes in complex networks. However, the study of multivariate interactions of nodes has become a trend in modern network science, which further promotes the research of hyperlink prediction. In this paper, we explore the impact of weights and the internal structure of hyperlinks on the accuracy of hyperlink prediction. By expanding upon traditional link prediction indices and fusing structural features such as hyperdegree, the degree of hyperlinks and so on, we develop new weighted similarity indices and propose a hyperlink prediction method based on weighted similarity. The simulation experimental and real dataset results show that our method enhances prediction accuracy in hyperlink prediction.