1 Introduction
Link prediction models are used to learn the distribution of links in graphs and predict the existence of potential links [1], [2], [3], [4]. In many real-world applications, the input data is represented as graphs, and link prediction models can be applied to tasks like friend recommendation in social networks [5], product recommendation in e-commerce [6], [7], [8], knowledge graph completion [9], protein interaction analysis [10], and metabolic network reconstruction [11].