Fusion-Based Knowledge for the Objective Force

Abstract

Introduction.Army Vision 2010 identifies information superiority as the key enabler for such force characteristics as dominant maneuver and precision engagement. These concepts are also central to the design and implementation of the Army's Future Combat System (FCS) and Objective Force. To establish and maintain information superiority, analysts and decision-makers need to identify, analyze and interpret pertinent information relative to achieving their task requirements. Currently, the sheer volume of information presented to Army intelligence analysts significantly exceeds their capabilities to fully analyze and interpret it in a timely manner. Consequently, the answers to commanders' critical information requirements (CCIRs) and priority intelligence requirements (PIRs) are typically based on a hasty, partial analysis of the information available. This condition of information overload experienced by analysts has the potential to significantly worsen for various reasons. First, our capabilities to collect, communicate and store data/information are steadily rising. Second, faster, more precise, and more lethal battlespace systems of the adversary cause an increase in operational tempo, as well as an increased risk to one's own forces, thereby resulting in more severe time constraints on analysis and decision-making. The nature of the analytical and interpretive tasks required to answer PIRs, and our ability to explain and justify their derivation, have largely been outside the realm of current machine capabilities. In recent years, a number of technologies and approaches have been developed (or matured) that show promise for addressing some of the key sources of difficulty characterizing this set of complex tasks either by emulating human methods or by providing automated support for aspects of these tasks that strain or exceed human cognitive capacities.To address this set of complex military intelligence problems, the U.S. Army Communications-Electronics Command and the U.S. Army Research Laboratory have submitted a collaborative proposal that would be carried out under the Army's Science and Technology Objective Program starting in FY03. One perspective for viewing this project is the Joint Directors of Laboratories (JDL) Data Fusion Model (Steinberg et al., 1998). With respect to this model, the present project will focus on problems associated primarily with data fusion Levels 2 and 3. However, it is our belief that data fusion problems are more likely to be understood and solved if they are approached more holistically by utilizing data fusion at any or all levels, if appropriate, to help solve a problem on a given level. The present paper provides a description of the technical challenges facing this project, and our current views on addressing them. The remainder of this paper begins by sketching the intelligence cycle and the military decision making process. Next, we discuss operational problems this project will address. This is followed by a description of some of the approaches and technologies...

Cognitive Complexities Impacting Army Intelligence Analysis

“…There are also bottom-up processes for interpreting and integrating information coming into the operations center. This use of both top-down and bottom-up processes has been observed in Army analysts in practice and characterized as highly similar to the blackboard model of problem-solving (Powell & Broome, 2002). Information includes reports returned in response to specific requests, and information volunteered from higher echelons, sister units, or combat units in the field.…”

“…In medicine, this initial stage results in a differential diagnosis of mutually exclusive explanations for a set of presented symptoms. Verbal protocols collected from experienced Army analysts suggest they form a limited set of hypotheses which function as a "differential diagnosis" of the battlefield when performing situation development (Powell & Broome, 2002) and that, more generally, the tasks carried out by these analysts exhibited a strong similarity to elementary task types characterizing diagnostic reasoning (Powell, 2004a). Current situation understanding and hypotheses guide direction of attention and determine what information is sought.…”

Supporting Intelligence Analysts with a Trust-Based Question-Answering System

“…One of the bigger challenges nowadays is the incapability of analysts to cope with (large amounts of) uncertain, ambiguous and complex information. Also several solutions to support analysts in dealing with this have been proposed [9], [10], [11] and [12]. However, it remains unclear what the added value of these proposed systems is.…”

“…The possibility of information overload also is a challenge for intelligence analysts [14]. According to [9] approximately 10,000 messages per hour are received and only 15,000 messages can be scanned a day [10]. [15] states that heavy information load affects performance, measured as accuracy or speed, negatively.…”

Issues and Challenges in Situation Assessment (Level 2 Fusion)

“…These displays show spatial and temporal relations from which to act. The display technology is domain dependent and requires operational considerations [30][31]. What follows is a small subset of key issues and challenges in knowledge representation and reasoning methods for Level 2 fusion in this task domain.…”