Multi-Graph Explorer: A Framework for Advanced Multi-Graph Analysis and Method Development

Yorgos Tsitsikas, Evangelos E. Papalexakis

Abstract: In recent years, multi-graphs have been gaining increasing popularity due to their ability to better capture the multi-faceted information of real-world graphs. This, in turn, has consistently provided superior insights and performance in related machine learning tasks. However, the analysis of real-world multi-graphs and the development of multi-graph methods is currently stifled by a few limitations. On one side, researchers often struggle to properly evaluate the performance of multi-graph methods they design due to a lack of high-quality benchmarks, but also a lack of tools that allow for efficient and seamless experimentation. On the other side, practitioners aiming to analyze real-world multi-graphs often struggle obtaining robust insights due to a lack of high-quality multi-graph methods. To this end, we present Multi-Graph Explorer: a MATLAB software designed to offer a user-friendly yet comprehensive, flexible, and extensible workflow for multi-graph analysis, aiming to break these barriers and accelerate progress in machine learning tasks involving multi-graphs.