Xiaohan Zhao     
Dr. Xiaohan Zhao

Postdoctoral Researcher
University of California, Santa Barabara
"Analyzing and processing real graphs”
Monday, February 23, 4:00 PM
Packard Lab Room 466

Abstract: As fundamental abstractions of network structure, graphs are everywhere, including Internet topologies, biological networks, and social networks. Understanding the structures of these graphs can provide insights on fundamental processes in these networks, such as information propagation, community formation and network dynamic models. In the past, most studies have focused on synthetic graphs or real graphs of limited size. Recently, however, large, real graphs from social and computer networks have become increasingly available, and they bring two new challenges to researchers working in related areas. First, the new wave of graph datasets are much larger than prior graphs, posing new challenges to query processing. Second, structures of these graphs are very different from existing graphs in prior studies, and their detailed analysis and modeling bring forth challenges in data privacy.

In this talk, I will describe two of my projects that target challenges in dealing with today's large, real graphs: efficiently computing random-walk distances on large graphs, and preserving graph privacy in sharing real graphs. First, I will introduce my work on near-realtime estimation of random-walk distances. My prior work proposed graph coordinate systems to compute node distance queries. I describe my new work on adapting these geometric embeddings to capture asymmetric distances such as hitting time and Personalized Pagerank. I show that using embedded graphs, random walks can be accurately estimated in microseconds instead of minutes, and the results reveal interesting insights on the impact of graph structure on random walk distances. Second, I will also describe some of my work on graph privacy, including a differentially-private graph model for privacy-preserving graph sharing.

Bio: Dr. Xiaohan Zhao is currently a Postdoc in Computer Science at University of California, Santa Barbara. She obtained BE and MS degree in Electronic Engineering at Tsinghua University, China, and received her PhD degree from UCSB last December. She is interested in big data mining and analysis, and data privacy. Her recent research work focuses on analyzing and processing large complex networks including online social networks, crowdsourcing networks, and location-based social networks. She also works on privacy and security problems in large complex networks.

© 2014-2016 Computer Science and Engineering, P.C. Rossin College of Engineering & Applied Science, Lehigh University, Bethlehem PA 18015.