Identifying Key Actors in Heterogeneous Networks
Abstract
The specific aim of this project is to advance and reconcile formal representations of social relations and coalition formation to improve accuracy for detecting significant or ÒpowerfulÓ actors within a total actor space when both resource and structural constraints must be taken into account, a highly probable situation in many operational contexts. The project will produce a new analytical capability forged by integrating the structural constraints that can be modeled by social network analysis (SNA) with the assessment of actors in contributing to collaborative coalitions modeled by game theory (GT). Relying on current formulations of these two well-accepted mathematical models used in isolation has in the past and will continue to misidentify important actors, focusing attention and resources off important targets and, in many instances, squandering good will. SNA by its very construction focuses on dyadic relations. And while there has been good work on inferring ÒrolesÓ from network metrics, these studies do not treat such roles as resources that must coalesce in order to achieve a desired outcome. Cooperative game theory models these types of coalitional relationships; however these models assume all actors can transact with any other, an assumption that is unreasonable in many operational settings. Moreover, despite its name, cooperative game theory models actors who cooperate only to gain the best individual advantage. While the other actors are necessary to achieving this advantage, they are not true allies. The situations we wish to model in this project specifically entail a set of allied actors seeking to achieve an advantageous outc ome, often at the expense of others outside the group. Finally, GT was designed to model small sets of actors, much fewer than is required to meet operational needs. To overcome these limitations requires a highly innovative synthetic methodology to reconcile incompatibilities and leverage complementary constructs from SNA and GT. We intend an iterative approach to the problem of integrating GT and SNA by (1) developing formal models, (2) instantiating the models into software, and (3) testing the accuracy of the models in terms of specificity and sensitivity on theoretically distinct network types to evaluate the reliability and get a deeper understanding of the limitations of the approach.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- May 07, 2018
- Source ID
- W911NF1710009
Entities
People
- Elisa Jayne Bienenstock
Organizations
- Arizona State University
- Army Contracting Command
- United States Army