Balanced Influence Maximization in the Presence of Homophily

Md Sanzeed Anwar, Martin Saveski, and Deb Roy

This repository contains code and data that can be used to replicate the analysis in the paper, "Balanced Influence Maximization in the Presence of Homophily" published in WSDM'21.

Guide to Files

models: contains the source code for network generation and seeding

• models/net_gen: contains the network generator models
  • models/net_gen/homophilic_net_gen.py implements the homophilic network generation model (Sec. 3.2)
  • models/net_gen/simple_net_gen.py implements the non-homophilic network generation model (Sec. 3.1)
• models/seeding: contains the seeding models
  • models/seeding/vanilla_greedy.py implements the model for the vanilla greedy influence maximization algorithm
  • models/seeding/balanced_greedy.py implements the model for the balanced influence maximization algorithm proposed in the paper (Sec. 4.1)
  • models/seeding/degree_threshold.py implements the model for the baseline algorithm by Stoica et al. (Sec. 4.3: Baseline)

raw_data/twitter: contains the four real-world networks used in the performance analysis of our proposed algorithm

experiments: contains scripts for all figures in the paper. experiments/fig_x corresponds to figure x in the paper, where x = 1,2,...,7

• experiments/fig_x/run_figx.py runs the experiment corresponding to figure x
• experiments/fig_x/get_csv.py generates a .csv file from the raw result data
• The scripts in experiments/fig_6 process the output our algorithm, while the scripts in experiments/fig_7 process the output of the baseline algorithm

plots: contains both the .csv files containing the experiment results demonstrated in our paper, as well as the plotting code needed to generate the figures in our paper.

To generate the each plot, run experiments/fig_x/run_figx.py which outputs the results in a .json file, then run experiments/fig_x/get_csv.py to convert them in the appropriate .csv format, and finally copy .csv file to plots/_csvs and run the corresponding script in plots.

Cite as

@inproceedings{anwar2021balanced,
    title={Balanced Influence Maximization in the Presence of Homophily},
    author={Anwar, Md Sanzeed and Saveski, Martin and Roy, Deb},
    year={2021}
    publisher = {Association for Computing Machinery},
    booktitle={Proceedings of the 14th ACM International Conference on Web Search and Data Mining},
    series = {WSDM '21}
}

License

This analysis code is licensed under the MIT license found in the LICENSE file.