Prediction of B-coupling in the Bipartite hypergraph through link prediction in the unimodal graph

Abstract

In the bipartite graph, B-coupling is defined as follow. Two nodes in one node set are B-coupled if they are connected to one or more nodes in the other node set in common. A node to which B-coupled nodes are connected in common is called the coupling node about the nodes. In this paper, we use the link prediction approach to predict the likelihood of a future B-coupling between nodes in one node set in the bipartite graph. The approach which will be addressed in this paper is the problem of identifying links in the unimodal graph obtained by the reduction of a bipartite graph over one of its node sets. The prediction of B-coupling in the bipartite graph can be mapped to the link prediction in the converted unimodal graph. In many real-world applications represented by the bipartite graph, the hyperedge can be considered. In this paper, the hyperedge is considered in the link prediction in the unimodal graph for predicting B-coupling in the intial bipartite graph. And for some applications it is necessary to find which nodes are coupling nodes. In this paper, we suggest the methodology which recommends coupling nodes about the predicted links after the link prediction process. By experiments, we show our work has better performances than previous works in the link prediction. And we also show that the quality of recommendation on coupling nodes is remarkable by several figures.