Can Cascades be Predicted?

  • Jure Leskovec, Stanford University; Paul Bennett, Microsoft

Social networks play a central role in spreading of information, ideas, behaviors, and products. As such “contagions” diffuse from a person to person they may go “viral,” and large cascades can form. However, a growing body of research has argued that virality and cascades may be inherently unpredictable. Thus, one of the central questions is whether information cascades can be predicted and possibly even engineered. In this talk, I will discuss a framework for predicting cascades and making them go viral. We study large sample of cascades on Facebook and find strong performance in predicting whether a cascade will continue to grow in the future. The models we develop help us understand how to create viral social media content: by using the right title, for the right community, at the right time.

Speaker Details

Jure Leskovec http://cs.stanford.edu/~jure is assistant professor of Computer Science at Stanford University. His research focuses on mining large social and information networks. Problems he investigates are motivated by large scale data, the Web and on-line media. This research has won several awards including a Microsoft Research Faculty Fellowship, the Alfred P. Sloan Fellowship and numerous best paper awards. Leskovec received his bachelor’s degree in computer science from University of Ljubljana, Slovenia, and his PhD in in machine learning from the Carnegie Mellon University and postdoctoral training at Cornell University. You can follow him on Twitter @jure http://www.x.com/jure

Series: Microsoft Research Talks