Classifying Types of Network Communities Using Motifs
Dave Braines recently presented to the Machine Learning, Data Analytics and Modeling - DATAM18 at CCS18 conference in Thessaloniki, Greece. Dave is a PhD student on the DAIS ITA project “Evolution of Human Systems” with IBM Research UK. His paper was titled Classifying Types of Network Communities Using Motifs.
Network motifs and graphlets are small sub-networks that occur within larger graphs sub as those that define an online network community or the actions taken within online networks (such as friendship, comments etc).
In the paper the authors outline a new approach to online network community classification which uses these motifs as the input to a feature vector classification technique which performs very well in identifying the type of network or community based on the structure that is observed. The presentation outlined the approach taken, the datasets used and the results obtained, along with a comparison against current state-of-the-art techniques in this area.
This work is part of the DAIS ITA project at CSRI and is a collaboration between University of Massachusetts Amherst, Cardiff University and IBM Research UK.