SLRIC package

SLRIC is a Python package for influence assessment in networks using short-range interaction centrality (SRIC) and long-range interaction centrality (LRIC) indices.

• Nodes 2 and 4 have different thresholds of influence (q);
• Nodes 1 and 2 influence node 4 as a group. Even though node 3 is connected to node 4, it does not influence it directly as node 3 is not a pivotal member in any group;
• Node 3 influences node 4 via node 2.

Website: https://github.com/SergSHV/slric

Authors: Fuad Aleskerov, Natalia Meshcheryakova, Sergey Shvydun (HSE University, ICS RAS)

Reference: Aleskerov, F., Shvydun, S., & Meshcheryakova, N. (2021). New Centrality Measures in Networks: How to Take into Account the Parameters of the Nodes and Group Influence of Nodes to Nodes (1st ed.). Chapman and Hall/CRC. https://doi.org/10.1201/9781003203421

Installation

Install the package from PyPI:

$ pip install slric

You can also install the latest version from GitHub:

$ pip install git+https://github.com/SergSHV/slric.git

Load SLRIC package:

>>> import slric

SRIC/LRIC Calculation for Simple Example

Generate a network using NetworkX package

>>> import networkx as nx
>>> G = nx.DiGraph()
>>> G.add_edge(1, 4, weight=7)
>>> G.add_edge(2, 4, weight=5)
>>> G.add_edge(3, 4, weight=2)
>>> G.add_edge(3, 2, weight=6) 

Case 1

• q=60% of weighted in-degree (in percentage);
• nodes have the same size (size = 1).

>>> slric.sric(G, q=60, size=1) # SRIC 
{1: 0.36363636363636365, 4: 0.0, 2: 0.09740259740259741, 3: 0.538961038961039}

>>> slric.lric(G, q=60, size=1, models='max') # LRIC (Max) 
{1: 0.27450980392156865, 4: 0.0, 2: 0.1470588235294118, 3: 0.5784313725490197}

>>> slric.lric(G, q=60, size=1, models='maxmin') # LRIC (MaxMin)
{1: 0.27450980392156865, 4: 0.0, 2: 0.1470588235294118, 3: 0.5784313725490197}

>>> slric.lric(G, q=60, size=1, models='pagerank') # LRIC (PageRank)
{1: 0.32165639923246203, 4: 0.0, 2: 0.18808528619697315, 3: 0.49025831457056473}

Case 2

• q=5 for each node (defined quota, dq);
• nodes have the same size (size = 1).

>>> slric.sric(G, dq=5, size=1) # SRIC
{1: 0.21153846153846154, 4: 0.0, 2: 0.28846153846153844, 3: 0.5}

>>> slric.lric(G, dq=5, size=1, models='max') # LRIC (Max)
{1: 0.25, 4: 0.0, 2: 0.25, 3: 0.5}

Case 3

• q=5 for node 2, q=10 for node 4;
• nodes have the same size (size = 1).

>>> d = dict()
>>> d[2] = 5
>>> d[4] = 10

>>> slric.sric(G, dq=d, size=1) # SRIC
{1: 0.2916666666666667, 4: 0.0, 2: 0.20833333333333337, 3: 0.5}

>>> slric.lric(G, dq=d, size=1, models='max') # LRIC (Max)
{1: 0.24137931034482757, 4: 0.0, 2: 0.17241379310344826, 3: 0.5862068965517241}

Write LRIC results to file

>>> from slric import lric, GraphQW
>>> ranking, lric_graph = lric(G, q=60, size=1, models=['max', 'maxmin'], data=TRUE)
>>> GraphQW.write_centrality(lric_graph, 'output.txt', separator=';', mode='w')

Additional features

1. If nodes size (size) is not defined, then size = weighted out-degree;
2. Similarly to threshold of influence (q), nodes size can be of dict() type;
3. Maximal group size can be limited using 'group_size' parameter (by default, group_size=4);
4. Maximal indirect influence limit can be defined using 'limpath' parameter (by default, limpath=3);
5. If LRIC version (models) is not defined, then LRIC (Max) is calculated by default (models='max' ).

License

BSD 3-Clause License

Copyright (c) 2019. Fuad Aleskerov, Natalia Meshcheryakova, Sergey Shvydun.

All rights reserved.