Multi-Graph Explorer

A Framework for Advanced Multi-Graph Analysis and Method Development

Overview

Multi-Graph Explorer is a MATLAB software that aims to enable researchers to streamline all stages of multi-graph method development, while also providing practitioners with a unified toolkit for efficiently discovering structure in real-world multi-graphs.

This project is associated with the following demo paper:

Yorgos Tsitsikas and Evangelos E. Papalexakis. Multi-graph explorer: A framework for advanced multi-graph analysis and method development. In Proceedings of the 33nd ACM International Conference on Information and Knowledge Management, October 2024

Also, the peer-reviewed version of the manuscript can be found here: https://www.cs.ucr.edu/~epapalex/papers/24-CIKM-Multi-Graph-Explorer.pdf

BibTeX users can cite this work as follows:

@inproceedings{multi-graph-explorer,
  title={Multi-Graph Explorer: A framework for Advanced Multi-Graph Analysis and Method Development},
  author={Tsitsikas, Yorgos and Papalexakis, Evangelos E.},
  booktitle={Proceedings of the 33rd ACM International Conference on Information and Knowledge Management},
  year={2024},
  month={October},
  doi={10.1145/3627673.3679213},
  isbn={979-8-4007-0436-9/24/10}
}

Features

• Artificial multi-graph generator: Generate artificial multi-graphs in a succinct yet flexible and comprehensive manner.
• Efficient non-standard parameter sweeps: Define value ranges for arbitrary combinations of all parameters and their subparameters for a thorough and systematic study of the multi-graph analysis workflow.
• Interactive graphical user interface: Adjust the parameters of the experiments and explore their results efficiently and intuitively.
• High-quality method implementations: Experiment with optimized implementations of existing multi-graph methods.
• Customization and Extensibility: Multi-Graph Explorer aims to maximize the ability of the users to customize the environment and to augment it with additional functionality.

multi-graph_explorer_demo_small.mp4

Dependencies

• MATLAB - Required (Tested on MATLAB 2019b)
• MATLAB Statistics and Machine Learning Toolbox - Required
• MATLAB Image Processing Toolbox - Required
• scikit-learn - Required (Tested with Python 3.6 and scikit-learn 0.24.2)
• MATLAB Parallel Computing Toolbox - Optional
• MTIMESX - Optional

Installation

• Make sure the current directory within MATLAB is the project's root directory at all times.
• For instructions on how to configure MATLAB to use Python see here.

Usage

• Run single_run_GUI.m to launch the Single-Combination Parameter Explorer.
• Run param_sweep_explore_demo.m to see an example of using the Multi-Combination Parameter Explorer.
• For further documentation, use MATLAB's help function (e.g. help single_run_GUI.m).

License

Apache License, Version 2.0