Award Date


Degree Type


Degree Name

Master of Science in Computer Science


Computer Science

First Committee Member

Justin Zhan

Second Committee Member

Yoohwan Kim

Third Committee Member

Juyeon Jo

Fourth Committee Member

Ge Lin Kan

Number of Pages



Human life is progressing with advancements in technology such as laptops, smart phones, high speed communication networks etc., which helps us by reducing load in doing our daily activities. For instance, one can chat, talk, make video calls with his/her friends instantly using social networking platforms such as Facebook, Twitter, Google+, WhatsApp etc. LinkedIn, Indeed, etc., connects employees with potential employers. The number of people using these applications are increasing day-by-day, and so is the amount of data generated from these applications. Processing such vast amounts of data, may require new techniques for gaining valuable insights. Network theory concepts form the core of such techniques that are designed to uncover valuable insights from large social network datasets.

Many interesting problems such as ranking top-K nodes and top-K communities that can effectively diffuse any given message into the network, restaurant recommendations, friendship recommendations on social networking websites, etc., can be addressed by using the concepts of network centrality. Network centrality measures such as In-degree centrality, Out-degree centrality, Eigen-vector centrality, Katz Broadcast centrality, Katz Receive centrality, and PageRank centrality etc., comes handy in solving these problems.

In this thesis, we propose different formulae for computing the strength for identifying top-K nodes and communities that can spread viral marketing messages into the network. The strength formulae are based on Katz Broadcast centrality, Resolvent matrix measure and Personalized PageRank measure. Moreover, the effects of intercommunity and intracommunity connectivity in ranking top-K communities are studied. Top-K nodes for spreading any message effectively into the network are determined by using Katz Broadcast centrality measure. Results obtained through this technique are compared with the top-K nodes obtained by using Degree centrality measure. We also studied the effects of varying α on the number of nodes in search space. In Algorithms 2 and 3, top-K communities are obtained by using Resolvent matrix and Personalized PageRank measure. Algorithm 2 results were studied by varying the parameter α.


Computer Sciences

File Format


Degree Grantor

University of Nevada, Las Vegas




IN COPYRIGHT. For more information about this rights statement, please visit