PI: COL John Graham and LTC Ian McCulloh (U.S. Military Academy)
In 2007, the Network Science Center began a research project sponsored by The United States Army Research Institute to examine new management tools for the information age. In particular, to what degree can email and blackberry communication data provide insight into informal networks within a military chain of command? In addition, what type of longitudinal behavior can be expected of military group dynamics? Will they continually evolve, or will they converge to a stable structure.
What was accomplished:
We obtained IRB approval to issue blackberries to every officer assigned to the Eisenhower Leadership Development Program (ELDP), and cadets serving in regimental chain of command. This provided a closed pool of leaders who had reason for daily email and phone interaction in the course of their duties. We had access to the formal organizational structure and could conduct surveys to understand the informal network structure of the participants.
Social network email data were collected over the course of the year-long program. The cadets in the chain of command were surveyed weekly during the Spring semester, reporting the time they spent with other group members, their ranking of friendship, their trust, and for select members, their geo-spatial data. The multiplex relations among the agents in the network were analyzed against each other and longitudinally.
Multi-agent simulation was used to model agents within an Infantry company and generate longitudinal social networks. Changes were introduced at known points in time in the simulation, in order to create a controlled environment to explore network change detection. Multiple change detection approaches were compared.
Methods for handling periodicity in data were developed using wavelets and the Fourier transform. In addition, change detection methods were automated in a social network package, ORA, under a grant from ARL.
Why it’s important:
Email provides insight into informal group dynamics. Email can be easily monitored to provide commanders information on the informal networks within their units. These networks can be monitored in real time. Network change detection algorithms have been developed and demonstrated to be effective in monitoring organizational behavior for significant change.
Methods for handling periodicity provide more accurate real-time monitoring of networks over time, with fewer false alarms.
Prior to organizations successfully carrying out some activity or event, there exists a change in the social network of the organization as members plan and resource the activity. Social network change detection may allow an analyst to detect the network change prior to the event, allowing a commander to get inside the decision cycle and influence the outcome.
This approach can also be used with adversarial network data to enhance intelligence of terrorist organizations.