An Empirical Method for Evaluating Robustness of Network Centrality Methods: The Case of "Dark Networks"
conference contributionposted on 25.02.2021, 00:35 by Joseph Shaheen
The Intelligence Community faces continuous challenges in the collection and aggregation of human-centric network data, partially due to a lack of availability of said data and tampering by the adversary. Moreover, chronic methodological issues manifest themselves in the applied analytical process post data retrieval; these can be summarized as a deficiency of empirical foundations.
This is especially true in the evaluation of node positions commonly captured by centrality measures. In this session, I present a simple bootstrapping method utilizing a scalar measure that provides a strong empirical foundation for determining centrality information loss in collected network data with minimal underlying assumptions. The method's use is illustrated using a theoretical network framework. Strengths, weaknesses, and opportunities are discussed.