Date of Award

5-2022

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Computer Engineering and Sciences

First Advisor

Chul-Ho Lee

Second Advisor

Rodrigo Mesa Arango

Third Advisor

Georgios Anagnostopoulos

Fourth Advisor

Marius Silaghi

Abstract

Large complex networks and graphs are everywhere. Examples include social networks, web graphs, communication networks, power grids, and transportation networks. The ongoing rapid expansion of such networks has triggered a tremendous amount of attention in various disciplines. Estimating the structural and topological properties of the large complex networks has been at the heart of the understanding of the networks. However, such estimations often become technically challenging. This dissertation is focused on how to estimate three non-trivial structural quantities on large networks efficiently and scalably by transcending the current state-of-the-art algorithms. The structural quantities to estimate are (1) a function of the eigenvalues of a matrix defined on a graph, (2) a distribution of set sizes defined on a graph when the information on the distribution is incomplete, and (3) how likely each node is vulnerable to malicious crawlers on a graph. We demonstrate not only the superiority of our proposed algorithms over the state-of-the-art algorithms but also their desirable properties such as computational efficiency and ease of implementation.

Download

Share

COinS