furkangursoy/comawarenetgrowth

Source code and datasets for "A Community-aware Network Growth Model for Synthetic Social Network Generation"

A Community-aware Network Growth Model for Synthetic Social Network Generation

This repository contains the source code and datasets employed in the following study.

---

If you find this useful for your research, please consider citing:

F. Gürsoy, B. Badur. "A Community-aware Network Growth Model for Synthetic Social Network Generation." Fifth International Management Information Systems Conference. 2018.

Full Text: https://arxiv.org/abs/1901.03629

@inproceedings{gursoy2018community,
  title={A Community-aware Network Growth Model for Synthetic Social Network Generation},
  author={G{\"u}rsoy, F and Badur, B},
  booktitle={Fifth International Management Information Systems Conference},
  year={2018},
  doi={10.6084/m9.figshare.7582082}
}

---

---

For a sample set of synthetic networks generated by the model, please visit http://furkangursoy.github.io/datasets.

File descriptions

ComAwareNetGrowth.R: the code file
karate.txt: karate network (appropriate references are given in the paper)
caltech.txt: caltech network (appropriate references are given in the paper)
input_orig.txt: input parameters for our different scenario experiments
input_karate.txt: input parameters for mimicking karate network
input_caltech.txt: input parameters for mimicking caltech network
output_(..).txt: output files corresponding to the input files

Input attributes

n: number of nodes
m: number of links
numberOfClusters: number of clusters (i.e., communities)
clusterProb: probabilities of belonging to each cluster
rp: a probability for making a random link
pp: a probability for making a link based on preferential attachment
c3p: a probability for making a link that closes a triangle
c4p: a probability for making a link that closes a quadrangle
comp: a probability for making the link within community

Note that, the probabilities described above are not actual probabilities but relative chances (automatically normalized by the R application when necessary). However, in the published paper, actual probabilities (i.e., those summing up to 1) are reported. Moreover, please briefly study the code to see how comp is multiplied with other four parameters to result in total of 8 probabilities for 8 link-making methods.

Output attributes

n: number of nodes
m: number of links cc: clustering coefficient
pl: average path length
mod: modularity
diam: diameter
alpha: power law exponent

Before you run the code

• Check & modify the names of input and output files. (Line 112 and Line 179)
• Set nOfTests to the number of test cases you have in your input file (or number of them you want to experiment with). (Line 111)
• Set the xmin parameter for fit_power_law() function to a desired value. (Line 169)
• Comment out Lines 173 - 176 to save the vertex and edge list of synthetically generated networks.

For any issues you might have, feel free to contact me at [email protected] or [email protected]

Feel free to use the generated networks or the source code, with appropriate references to the original paper.