Social network analysis is a necessary aspect of learning terrorist/criminal networks, or of how people interact through social networking sites like Facebook and Twitter. The metrics used to evaluate these networks are equally as important, centrality measures being the most common. Being the most routinely used metrics, centrality measures have been tested under various conditions. However, there is a lacking focus on systematic error applied to computer generated connected undirected small-world networks. The goal of this paper is to test the centrality measures (betweenness, closeness, and degree) in such networks and test their robustness after the introduction of systematic error. The experiment found that the metrics performed poorly and their use would have to be determined based on the situation and consequences.
展开▼
