Abstract
The phenomenon of terrorism is deemed one of the fundamental challenges in national security. Creating defensive technologies to mitigate terrorist attacks requires a simultaneous investigation of contextual relationships among their various dimensions. We proposed and evaluated a graph-based methodology to analyze terrorist networks through co-clustering in a multimode basis. Since there are many heterogeneous relationships in terrorist networks depending on the dimensions used during analysis, we utilized the clustering indicators of the multimode structure discovered in bi- and multimode graphs. Objects and activities that co-occur during terrorist attacks are identified by applying conventional clustering on those indicators. The novelty of our method is in the incremental creation of the multimode structure using its bi-mode counterparts. Our approach is evaluated using these measures: clustering stability and association confidence. The experimental results yields encouraging results in terms of simultaneous clustering of heterogeneous objects in terrorist networks.
Original language | English (US) |
---|---|
Pages (from-to) | 1053-1074 |
Number of pages | 22 |
Journal | Information Systems Frontiers |
Volume | 20 |
Issue number | 5 |
DOIs | |
State | Published - Oct 1 2018 |
Externally published | Yes |
Keywords
- Multimode clustering
- Singular value decomposition
- Social network analysis
- Terrorist networks
- k-means
ASJC Scopus subject areas
- Software
- Theoretical Computer Science
- Information Systems
- Computer Networks and Communications