Node Classification for Signed Social Networks Using Diffuse Interface Methods

MATLAB implementation of the paper:

P. Mercado, J. Bosch, and M. Stoll, Node Classification for Signed Social Networks Using Diffuse Interface Methods. In ECMLPKDD 2019.

Content:

• example.m : contains an easy example showing how to use the code

• sample_script_wikipedia.m : example on how to use our approach on Wikipedia Datasets from our paper

Usage:

Let Wpos,Wneg be adjacency matrices of positive and negative graphs, supervised_nodes_idx an array with indexes of labeled nodes, labels_of_supervised_nodes an array with the corresponding labels, Laplacian_str a string indicating which signed Laplacian to use, and numEigenvectors a scalar indicating how many eigenvectors to take.

Node Classification for signed graphs via diffuse interface methods is performed via:

Y_hat = NCSN_using_diffuse_interface_methods(Wpos, Wneg, supervised_nodes_idx, labels_of_supervised_nodes, Laplacian_str, numEigenvectors);

Quick Overview:

Citation:

@InProceedings{Mercado:2019:ecmlpkdd,
author="Mercado, Pedro and Bosch, Jessica and Stoll, Martin"
title="Node Classification for Signed Social Networks Using Diffuse Interface Methods",
booktitle="ECMLPKDD",
year="2019",
}