Large Graph Mining - Patterns and Tools

Invited talk to WBD
Christos Faloutsos
May 17, 2024

Abstract

Given a million-node graph (like, e.g., who-views-what-when), how can we find patterns, how can we do forecasting and how can we spot fraud and anomalies? Here we focus on three of the many tools that graph mining has to offer: 1) pattern discovery 2) belief propagation and 3) time series analysis and forecasting. For the first, we will cover old and new patterns that appear in large graphs (six degrees, power-laws), and show how they can help us spot anomalies. For the second, we will briefly describe the powerful method of 'belief propagation' and present some success stories. For the third, we will show how time analysis can help spot strange behaviors.

Slides


Last updated: May 17, 2024, by Christos Faloutsos