This repository contains Python implementation of team formation algorithms that were introduced in the last decade.
We proposed two algorithms that rely on power law property of degree distribution of a social network.
We named these Team formation based on Power Law algorithms as TPLRandom and TPLClosest algorithms.
DBLP dataset is a collection of 21 networks.

Description of files

• *.gml files are the networks
• clean_data.sh Bash script file to clean, extract information from dblp-2015XX.xml.gz file and generate all network files as given in appendix of the manuscript.
• *.m gnuplot scripts to generate plots of the results
• main_<algorithm>.py main file to run the team formation algorithm
• Team.py python module with team class. This class contains task and team and their representation as skills as member variables. Methods for finding the team size diameter, sum distance and leader distance and visualization of the team.
• Algorithms.py This module contains the implemetation of all team formationn algorithms used for comparison including proposed algorithms
• utilities.py This modules contains the utilities like conversion of list to frequency, get all k-hop neighbors,etc.

Running the code

1. Download dblp-yyyy-XX.xml.gz from dblp repository.
2. Execute clean_data.sh file specifying the details of downloaded file
3. Run main_<algorithm>.py file to for each algorithm #Plots generation
4. power_law_plots.m this file generates the powerlaw property by degree distribution of all the networks
5. tf_plots.m will plot the comparions of communication costs(diameter, LD and SD) to task size for all the networks.