An Evaluation of Algorithms for Centrality Analysis of Twitter Users.

摘要

Centrality measures of social network are important in various areas and applications; including companies that take interests in analyzing social networks. Institutions which utilize centrality find it both beneficial and rewarding. I present this paper as a master's thesis requirement as a comprehensive approach to "An Evaluation of Algorithms for Centrality Analysis of Twitter users". This thesis examines the algorithms of Floyd--Warshall, Johnson, and Brandes by applying them to a twitter data set. This thesis also provides performance evaluations to each of the three algorithms for betweenness centrality. In addition to providing a complete implementation for both Warshall and Johnson's algorithms in R environment, this thesis generates beneficial insight for businesses and organizations alike who want to know who is the central or influencer in a given specific network and which algorithm to use. Keywords: Social network analysis, Centrality analysis, Twitter, Floyd--Warshall algorithm, Johnson algorithm, Brandes' algorithm, R language.