Soar Technology has developed the technology to create exploratory models of social networks and organizations to understand their emergent properties, interactions, and behaviors. Such models are critical for exploring low-probability, high-risk situations in intelligence analysis and predictive modeling. For many types of exploratory research, there is no well-defined or realistic solution that can fully cover the problem space. Social and cultural modeling of complex organizations and situations is one such type of research where inherent dynamics and uncertainty prohibit analytic solutions.
The Problem
The intelligence community finds itself overwhelmed with information from multiple sources, and often a primary task is to find needles in haystacks – whether watching for potential terrorist operations or conflicts between nations. This job is particularly difficult when an intelligence target is a closed regime – a country or organization whose decision-making apparatus is not directly accessible for observation. The intelligence analyst must rely on open source reports that may be directly controlled by the regime or covert sources that may be unreliable at best. The intelligence process itself often (sometimes necessarily) includes biases that preclude the investigation of low-probability, high-impact scenarios. Either so much information must be sifted through that there is no time to investigate these areas, or incomplete or inaccurate information biases the analyst against exploring those possibilities. Furthermore, those biases are often only implicit in the analyst’s work products, and make the results of the analysis more difficult to place in context. This tacit knowledge in the mind of the analyst is very likely lost from corporate memory when that analyst retires.
Agent-Based Social NetworksAGILE fulfills the first role by means of agent-based social model of various aspects of the target countries. A country is represented by multiple intelligent agents, each with its own beliefs, desires, and intentions, cooperating as a decision-making team to solve their goals collectively. The agents weigh problems and potential solutions brought to the team’s attention through the lens of their own beliefs and goals, and contribute their views in terms of votes for or against actions. Each country can control aspects of a world model that embodies the “physics” of the game, including how economies work and constraints on military movements, etc. Countries can interact by manipulating aspects of the world model (changing tariffs, moving troops, etc.), or by proposing international agreements that might, in total, benefit both countries. Once the analyst has established initial world conditions, the agents start to work toward their goals, interacting in ways that may be unexpected. AGILE allows the user to specify a space of possible world conditions, including those that are possible but not currently true, to see how the agents might behave under different conditions, and to see what turns the relationships between countries might take under varying circumstances.
. ]
©2008 Soar Technology, Inc. All rights reserved
Legal Notice | Site Map
- Projects
- Publications
- Scientists
- Links