All source code of "Community Detection using Social Relations and Trajectories".

Requirements

• Unix-like system with the /tmp directory
• At least 8 GB of RAM
• clang++, supporting C++17, invokable with clang++
• The following C++ libraries installed:
  • Boost C++ Libraries
  • Eigen
  • ben-strasser/fast-cpp-csv-parser
  • p-ranav/argparse
  • pybind/pybind11
• Python 3, invokable with python3, with the following extensions:
  • folium
  • haversine
  • matplotlib
  • more-itertools
  • networkx
  • notebook (JupyterLab FAILS to display pyecharts visualizations!)
  • numpy
  • pandas
  • pybind11
  • pyecharts
  • scipy

Replication Instructions

NOTE: if you need to change experiment parameters, edit parameters.sh.

Extract Social Networks and Trajectories

unzip social_networks.zip
unzip trajectories.zip

Build Tools

Run bash build_tools.sh.

Run Experiments

Run bash run_experiments.sh. This process takes a long time (around a day).

Data Analysis

Run the following Jupyter Notebooks:

1. case_study.ipynb
2. average_trajectory_similarity_calculation_time.ipynb
3. spatial_temporal_distance_correlations.ipynb
4. m_values_to_sizes.ipynb
5. m_values_to_pairwise_distances.ipynb
6. m_values_to_pairwise_similarities.ipynb
7. community_detection_times.ipynb