SITIS Archives - Topic Details
Program:  SBIR
Topic Num:  OSD10-HS1 (OSD)
Title:  Decision Superiority through Enhanced Cultural Intelligence Forecasting
Research & Technical Areas:  Human Systems

Acquisition Program:  
  Objective:  The objective of this topic is to develop a decision support tool capable of visually presenting complex cultural and social attitudinal/behavioral variables extracted from large datasets that can be used to forecast attitudes and behaviors in a cultural context.
  Description:  Social network analysis is a valuable methodology for understanding relationships of social groups [1], tracing interactions among individuals [2], and describing data diffusion behavior in delay tolerant networks [3]. The growth of information available from on-line social networking sites and the availability of multi-source sensors has produced a corpus of data that, if sufficient data mining capabilities existed, could aid intelligence analysts in understanding terrorist intentions in near-real time [4]. Military leaders at tactical, operational, and strategic levels of command need to go beyond the typical social network analysis concepts in order to effectively operate in full spectrum operations. Specifically, the cultural intelligence component must be added to the commander’s decision support toolkit. Knowledge of sociocultural influences on attitudes and behaviors in a population is critical to effective interactions on the part of military personnel with local populations. As General (Retired) Anthony Zinni, U.S. Marine Corps, former U.S. Central Command Commander stated, "You have to understand the culture you're getting involved in. We never do a good job of culture intelligence, of understanding what makes people tick, what their structure is, where authority lies. Culture bias limits our ability to understand what is going on around us." [5] Sociocultural modeling techniques have progressed in recent years, as evidenced by the plethora of games built on a variety of sociocultural domains [6]. Several critical challenges have impeded the design of real-time decision support tools that deal with cultural factors. The first is the availability of cultural data from populations of interest. Frequently these populations are from nations that do not collect, store, and share data. Additionally, public trend data that would be obtained from public polling sites is difficult if these nations pose threats to data collectors. Finally, the ability to develop decision support tools that allow real-time forecasting predictions are difficult because the predictions are based on correlations between attitudes, influence, and behaviors of political, social, and religious leaders in the society. The interactions between these factors can be difficult to understand, model, and predict. However, military planners and decision makers require access to the forecasting capabilities of socio-cultural models, as well as the data necessary to populate such models. Gathering and interpreting relevant data can be a challenge when planners are far removed from the situation on the ground. In addition, interaction with advanced modeling capabilities must be structured in such a way as to provide support to the decision making process without placing an undue burden on the end user. Such a tool should include the following: 1.) the ability to collect data in real-time from the field that are directly relevant to modeling approaches; 2.) the ability to feed that data directly into existing models to produce reliable forecasts and accompanying measures of uncertainty; and 3.) the construction of support tools to allow model outputs to be translated directly into actionable decisions. Challenges for this topic include 1) identifying relevant data sets that provide sufficient sociocultural information to develop models, 2) identifying new social network associations that take into account sociocultural factors, 2) developing methods to extract features associated with social influence in groups, 3) defining and testing metrics for describing temporal aspects of social network behavior, and 4) identifying methods to predict novel concepts that arise from social behavior and attitudes.

  PHASE I: Define requirements for a decision support tool that would incorporate a sociocultural database for a population of interest. Prepare a proof of concept use case for how sociocultural data would be developed, modeled, and presented in a software application for use by military users. Prepare a literature review documenting methods, theoretical approaches, and software applications.
  PHASE II: Develop a software application that uses sociocultural modeling and forecasting for a population of interest. Demonstrate visual application of forecasting through the use of social network graphs of a relevant type. The prototype should lead to a demonstration of the capability. Test the prototype in an environment to demonstrate feasibility. Propose a verification and validation process.

  PHASE III: Produce a system capable of deployment in an operational setting of interest. Test the system in an operational setting in a stand-alone mode or as a component of larger system. The work should focus on capability required to achieve transition to program of record of one or more of the military Services. The system should provide metrics for performance assessment.

  References:  [1] Waskiewicz, T. & LaMonica, P. (2008). Developing an intelligence analysis process through social network analysis. In M. Blowers and Alex F. Sisti, Eds., Evolutionary and Bio-Inspired Computation: Theory and Applications II, Proceedings of SPIE Vol. 6964, 69640B. [2] Kahn, M. U. & Khan, S. A. (2009). Social networks identification and analysis using call detail records. ACM: ICIS, November 24-26, Seoul, Korea. [3] Zhang, Y. & Zhao, J. (2009). Social network analysis on data diffusion in delay tolerant networks. ACM: Proceedings of MobiHoc 09, May 18-21, New Orleans, LA. [4] Drozdova, K. & Samoilov, M. (2009). Predictive analysis of concealed social network activities based on communication technology choices: Early-warning detection of attack signals from terrorist organizations. Computational Math Organizational Theory, DOI 10.1007/s10588-009-9058-2. [5] Center for Army Lessons Learned. Agribusiness Development Teams (ADT) in Afghanistan Handbook 10-10, Nov 2009, Chapter 6. Available on the web at: [6] Stitts, K. B., Phillips, C. L., & Geddes, N. D. (2009). Validation of sociocultural models and meta-models via serious games. IEEE, 5/09.

Keywords:  sociocultural modeling, forecasting, social networks, sociocultural measures, decision support tools.

Additional Information, Corrections, References, etc:
Ref #1 - 6: Six (6) new Refs. 1-6 were added 4/21/10.

Questions and Answers:
Q: 1.) Should the decision-aid operate in real-time?
2.) Should the decision-aid run autonomously? In other words, should the decision-aid actively monitor the user and provide information without being prompted by the user?
A: 1. No, the decision aid does not have to operate in real time.
2. that would perhaps be a good capability to build toward, but that might be a step too far now. However, if it is possible now, with current technology, I think that would be a good feature.
Q: Your topic mentions "sociocultural" data but the problems addressed fall broadly under a host of disciplines loosely under Operations Other than War (e.g. see
Does your sociocultural paradigm fit within this general arena as well, or is it more specific – such as a subcomponent of ONR’s HSCB modeling program containing so-called social and cultural models (but not necessarily human or behavior-centric models)? Or, perhaps as the citations seem to hint, are you referring to sociocultural data as that which precipitate social network analyses specifically?

If I perhaps haven't quite worded my query appropriately, maybe you might point us toward some of the "open source online sites" for your sociocultural data which might help me better understand your view of data and models appropriate for proposed decision-aids.
A: We are taking an open view of sociocultural data at this point, and our focus is on the host nationals (as opposed to multinational team-members in a Combined Joint Task Force).

We are seeking to improve our ability to do social network analysis, so that is our ultimate goal. For example, if you knew that tribal affiliations in a region was more important (to some scale) than regional affiliations, links between individuals might be stronger in one case than in another.

Record: of