Simi Wang, University of North Carolina at Chapel Hill


Fast Algorithm to identify the communities in multi-slice network

Abstract: In my project, I improved the 2-D (network) Louvian Code into 3-D(Multi-slice network) using Peter Mucha's theory**, and gave the algorithm in bi-part, directed, signed Multi-slice network. I also apply my algorithm into some network in real world. Zachary Karate Club network, with 16 slices and 34 nodes in each slice only takes around 50 seconds. When considered the roll call voting in the US senate across time, it consists of 110 network slices and 8794 nodes in total, my code takes around 15 minutes. It can be seen that the if the adjacency and coupling matrix becomes more dense, my algorithm can be more efficient.

**In Peter Mucha's previous paper "Community Structure in Time-Dependent, Multi-scale, and Multiplex Networks"(Science 328, 876-8, 2010), he introduces a way to identify the communities of the multi-slice network through maximizing the modularity.

Mentor: Peter Mucha (University of North Carolina at Chapel Hill)