pdacorn / Finding-core-periphery-structure-with-node-influences

for paper Finding core-periphery structures with node influences

Finding core-periphery structure with node influences

This code implements the influence-based core-periphery detection algorithm (ICPA) in python for detecting multiple core-periphery pairs given adjacency networks. This work can be cited as follows:

Xin Shen, Sarah Aliko, Yue Han, Jeremy I Skipper, & Chengbin Peng. (2021). Finding core-periphery structures with node influences. IEEE Transactions on Network Science and Engineering, In press 10.1109/TNSE.2021.3138436.

Running influence-based core-periphery detection algorithm (ICPA) models

From Test.py, we can see how to use the uploaded files.

Firstly, you may create the adjacency matrix with the following approach.

Besides email-Eu-core-department-labels.txt, many other data sets are available, for example, at http://snap.stanford.edu/data

(see line 2-10 in Test.py)

import icpa_sparse_matrix
import timeit
import readNet
from importlib import reload
from scipy.sparse import csr_matrix

dt0=readNet.Edge2Sparse('email-Eu-core-department-labels.txt')
dt0_sparse = csr_matrix(dt0)

The returned dt0_sparse is a sparse adjacency matrix.

Secondly, you can obtain pair identities and core scores from our algorithm (ICPA) outputs

1. The parameter corenessnum indicates whether a node is a core or a periphery. Here, zero indicates a periphery node, and one indicates a core node. (see line 26 in Test.py)

corenessnum,corescore =icpa_sparse_matrix.step3(power, influenceratio = influenceratio)

2. The parameter of pairidnum indicates each node's pair id. (see line 20 in Test.py)

pairidnum = icpa_sparse_matrix.step2(power,topNinfluencer = 1)