Skip to content

Methods for identifying and ranking users who alternately behave as contributors and as lurkers over multiple layers of a multilayer social network

Notifications You must be signed in to change notification settings

andreatagarelli/mlALCR

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Multilayer Alternate Lurker-Contributor Ranking (mlALCR)

Example multilayer social network, which illustrates eighteen users located over three layer networks.

Copyright 2017-2018 Diego Perna, Andrea Tagarelli, DIMES Dept., University of Calabria, Italy

mlALCR is a Python software for identifying and ranking users who alternately behave as contributors and as lurkers over multiple layers of a multilayer social network. The software, in addition to an implementation of the mlALCR method, contains the implementation of two baselines and an extension of LurkerRank to multilayer networks. The implemented methods were originally defined and described in the peer-reviewed scientific publication reported below.


TERMS OF USAGE: The following paper should be cited in any research product whose findings are based on the code here distributed:

Diego Perna, Roberto Interdonato, Andrea Tagarelli.
Identifying Users With Alternate Behaviors of Lurking and Active Participation in Multilayer Social Networks.
IEEE Trans. Comput. Social Systems 5(1): 46-63 (2018).


Instructions

The relative path of the working directory is assumed to be ./data/ which must contain the input multilayer network(s). Each multilayer network is stored into a text file with extension .ncol modeling the edge lists over the layers. If the working directory contains more than one multilayer network, the input files of the networks will be processed sequentially, according to lexicographic order of the file names.


The format of the input .ncol files:

Each row in a .ncol file contains a triple of values (integer, integer, string) corresponding to the IDs of source-node, target-node, and layer, respectively, separated by ;. Moreover, the .ncol file is required to have the first row corresponding to a header (e.g., source;target;layer).

Options

The software runs one of the following ranking methods, by specifying the option --method $method_name:

  1. mlALCR
  2. mlLR
  3. LR (LurkerRank)
  4. Baselines (LRa1, LRa2)

The software also evaluates the correlation between two ranking solutions, by specifying the option --eval $method_name:

  1. Kendall (Kendall Tau rank correlation coefficient)
  2. Fagin (Fagin's intersection metric --- it requires to specify the number k of top users to be selected)

Usage Examples

  1. Ranking methods:
  • Run mlALCR on all the input files in the working directory, with alpha1=0.5 and alpha2=0.95:
    python3 mlALCR_Project.py --method mlALCR --alphas 0.5,0.95

  • Run LurkerRank on all the input files in the working directory, with alpha=0.75:
    python3 mlALCR_Project.py --method LR --alpha 0.75

  1. Evaluation methods:
  • Compute the Kendall Tau rank correlation coefficient between ranking1.txt and ranking2.txt:
    python3 mlALCR_Project.py --eval kendall --f1 ranking1.txt -f2 ranking2.txt

  • Compute the Fagin's intersection metric between ranking1.txt and ranking2.txt, with k=100:
    python3 mlALCR_Project.py --eval fagin -k 100 --f1 ranking1.txt -f2 ranking2.txt