Application of Confidence Intervals to Text-Based Social Network Construction
Abstract
With the increasing importance of gathering intelligence on insurgent and terrorist groups, social network analysis (SNA) has become an important analytic tool, SNA is the mathematical methology of quantifying connections between individuals and groups. This research is focused on the concept of centrality, which is a mathematical process of determining which node in a network is the most central, or connected. Thematic, or intangible, relationships consist of entities that are not directly connected, but who share similar ideologies. While the concept of centrality - the most connected node - remains the same, the question becomes how to determine if two nodes are connected where a tangible relationship is not present To determine if there is a connection, t-confidence intervals are constructed for each entity. If they share overlapping confidence intervals, they are connected. The connection is weighted based on a scaled difference between the means of the confidence intervals. The final network consists of nodes connected only across all of the chosen intangible or thematic fields, Analysis of the network is conducted using measures of degree centrality. This paper proposes a new algorithm for determining connection between nodes in a thematic network, using an analysis of radical jihadist writings to demonstrate the applicability of the method.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 2006
- Accession Number
- ADA490625
Entities
People
- Ian Mcculloh
- John Graham
- Julie Paynter
Organizations
- United States Military Academy