Faculty Contact: Ambuj Singh
Abstract: Understanding and modeling complex network processes—such as information cascades in online social networks—is an important task in many real-world applications. With the enormous amount of data generated on the Internet today, it is easy for us to get lost in information overload and fail to see the big picture of how these network processes happen. Can we summarize the spread of information in social networks by a small yet interpretable set of cascading subgraphs, each of which represents a set of connected users frequently participating in the same network processes? In our project, we aim to solve this problem for large-scale social networks and formulate it as a Binary Matrix Factorization with a network constraint. Showing that the problem is NP-hard, we further propose a greedy approximate algorithm as well as two scalable variants of this problem.
- Winter 2017: Furkan Kocayusufuglu and Minh Hoang