Counterterrorism Mining for Individuals Semantically-Similar to Watchlist Members
A key counterterrorism problem is how to identify people that should be added to a watchlist even though they have no direct communication with its members. One of the main ways a watchlist is expanded is by monitoring the emergence of new persons who est
- PDF / 552,890 Bytes
- 25 Pages / 439.37 x 666.142 pts Page_size
- 53 Downloads / 153 Views
Abstract A key counterterrorism problem is how to identify people that should be added to a watchlist even though they have no direct communication with its members. One of the main ways a watchlist is expanded is by monitoring the emergence of new persons who establish contact with those on the list. Unfortunately, this severely limits the time horizon for managing risks of dark network behaviors because individuals are already actively involved with one another and more likely to be discussing and planning terrorist actions. In contrast, a wider time horizon results from identifying individuals who do not yet have communication with watchlist members, while they have highly similar semantic networks. Discussion forums are considered a primary source of intelligence about plans for dark behaviors. The research reported here develops a method for locating individuals in discussion forums who have highly similar semantic networks to some reference network, either based on watchlist members’ observed message content or based on other standards such as radical jihadists’ semantic networks extracted from messages they disseminate on the internet. This research demonstrates such methods using a Pakistani discussion forum with diverse content. Of those pairs of individuals with highly-similar semantic networks, 61% have no direct contact in the forum. It is likely that adding to watchlists individuals who have a high match to a reference semantic network lengthens the time horizon for identifying high risk dark behaviors.
J.A. Danowski () Department of Communication, University of Illinois at Chicago, Chicago, IL, USA e-mail: [email protected] U.K. Wiil (ed.), Counterterrorism and Open Source Intelligence, Lecture Notes in Social Networks 2, DOI 10.1007/978-3-7091-0388-3 12, © Springer-Verlag/Wien 2011
223
224
J.A. Danowski
1 Introduction 1.1 Chapter Focus: Counterterrorism Text Mining to Find Similar Semantic Networks The focus of this chapter is on text analysis using methods of social network analysis. Before more fully conceptualizing semantic networks, I highlight the counterterrorism strategy to which semantic network analysis will be adapted. One probably has a good sense of network analysis to facilitate comprehension of basic ideas framing this chapter. The motivation for the research reported here came from a talk at the European Intelligence and Security Informatics 2008 conference in Esbjerg, Denmark. It was not my talk but an intelligence analyst’s. Later in this chapter I, will tie together the nature of my counterterrorism paper with the more recent evolution of my research. I was there to report on research applying my optimal message generation software, OptiComm [20] to a corpus of approximately 8,000 international news stories about Al Qaeda. The objective was to explore the formulation of messages that might be appropriate for a counterterrorism public information campaign. In the aggregate word network I identified the shortest paths across the stories between the words ‘Al Qaeda’ and ‘bad.’ [19].
Data Loading...
