cidre 0.0.0

Algorithm for finding anomalous groups in networks

Python package for the CItation-Donor-REcipient (CIDRE) algorithm.

Please cite:

Sadamori Kojaku, Giacomo Livan, Naoki Masuda. Detecting citation cartels in journal networks. arXiv:2009.09097 (2020)

Install

pip install cidre

Minimal example

import cidre

alg = cidre.Cidre(group_membership)
groups = alg.detect(A, threshold = 0.15)

• group_membership: If the network has communities, and the communities are not anomalous, tell the community membership to CIDRE through this argument, where group_membership[i] indicates the group to which node i belongs. Otherwise, set group_membership=None.
• A: Adjacency matrix of the input network (can be weighted and directed). nx.Graph or scipy.sparse_matrix. if scipy.sparse_matrix format, A[i,j] indicates the weight of edge from node i to j.
• threshold: Threshold for the donor and recipient nodes. A larger threshold will yield tighter and smaller groups
• groups: Detected groups. This is a list of special class, Group.

The donors of group can be obtained by

groups[0].donors # {node_id: donor_score}

• group.donors is a dict object, with keys and values corresponding to the node ID and the donor score.

Similarly, the recipients of a group together with their recipient scores are given by

group.recipient # {node_id: recipient_score}

Visualization

ax = plt.gca()
dc = cidre.DrawCartel()
dc.draw(group, ax = ax)

The labels beside the nodes are the ID of the nodes, or equivalently row ids of the adjacency matrix A.

To put the node labels, make a dictionary from the ID to label, like node_labels = {0:"name", 1:"name 2"}, and pass it by node_labels = node_labels.

Example script

• Toy network with communities
• Citation network of journals in 2013