Paper: Vertex clustering in diverse dynamic networks

Here at Level, we endeavor to use cutting edge data science to solve our business problems. We also believe in contributing back to the scientific community where we can: we actively conduct and publish fundamental research, attend conferences like the JMM, and maintain active collaborations within academia and beyond. In that spirit, we're delighted to announce our new peer-reviewed paper Vertex clustering in diverse dynamic networks, which was just accepted for publication in the open-access journal PLOS Complex Systems. This paper is a joint work with Dr. Olga Dorabiala, who was a post-doctoral fellow at the University of Washington during this project.

This paper extends our previous research on finding natural clusters within dynamic networks. In static networks, vertex clustering is an important—and largely solved—problem with applications ranging from social networks to satellite communication networks. In the dynamic setting, where the network is constantly changing, defining and finding clusters is much harder. With our collaborators, we developed STGkM, a novel method for clustering in dynamic networks. In this new paper, we demonstrate that it is not only theoretically sound, but that it also has state-of-the-art performance on several important benchmark datasets. We provide a detailed description of the algorithm, prove three new theorems about our algorithm, and run several experiments on large dynamic network datasets. All of the code and data for the paper are publicly available on Github.

You can access the pre-print here:
https://docsend.com/view/a6nkjan6y6istxk7

We will update this post with the final version of the paper later this year.