Abstract
Graph embedding is gaining its popularity for link prediction in complex networks and achieving excellent performance. However, limited work has been done in sparse networks that represent most of real networks. In this paper, we propose a model, Sparse Structural Network Embedding (SSNE), to obtain node representation for link predication in sparse networks. The SSNE first transforms the adjacency matrix into the Sum of Normalized $H$-order Adjacency Matrix (SNHAM), and then maps the SNHAM matrix into a $d$-dimensional feature matrix for node representation via a neural network model. The mapping operation is proved to be an equivalent variation of singular value decomposition. Finally, we calculate nodal similarities for link prediction based on such feature matrix. By extensive testing experiments bases on synthetic and real sparse network, we show that the proposed method presents better link prediction performance in comparison of those of structural similarity indexes, matrix optimization and other graph embedding models.
Highlights
IN natural complex systems, there are many entities, which interact with each other in a complicated way
The experimental results based on sparse networks show that the link prediction method based on Sparse Structural Network Embedding (SSNE) outperforms other methods based on structural similarity indexes, matrix optimization, and other graph embedding models
As graph embedding is recently used for link prediction in complex networks, this paper proposes a novel link prediction method based on SSNE constructed in the framework of graph embedding
Summary
IN natural complex systems, there are many entities, which interact with each other in a complicated way. Chen et al.: SSNE: Effective Node Representation for Link Prediction in Sparse Networks same time, keep the structure feature and inherent attribute of the graph [18]–[22]. The link prediction method based on DeepWalk is shown to predict better the possible incidence of MicroRNA genetic disease [24], [25], as well as individual multiple interests or attributes [26], [27] These embedding models succeed in link prediction in many natural networks, they involve critical experience-tuned parameters, such as the sampling length of a random walk and the number of random walks [23]. The experimental results based on sparse networks show that the link prediction method based on SSNE outperforms other methods based on structural similarity indexes, matrix optimization, and other graph embedding models.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
