Recommendation as link prediction in bipartite graphs: A graph kernel-based machine learning approach

► We propose a kernel-based approach for link prediction and recommendation. ► We design a graph kernel to exploit features in the context of focal user-item pair. ► The kernel works with a one-class SVM algorithm to predict user-item interactions. ► We prove the validity and computational efficiency of the graph kernel. ► Our model outperforms benchmarks, particularly for large amounts of recommendations.