Army Intelligence Analysis and Interpretation: Assessing the Utility and Limitations of Computational Diagnostic Reasoning

摘要

The U.S. Army's Future Force is critically dependent on information superiority levels that will support timely, quality decision-making during high tempo operations. The Future Force is anticipated to produce unprecedented levels of data requiring analysis. Data fusion is considered by some as the potential solution to handling this severe data overload. The Joint Directors of Laboratories Data Fusion model categorizes data fusion-related functions at a high level of generality. For fusion in the Army, little has been published reflecting an elaboration of functionality on levels 2 and 3 of this model, both of which are viewed as critical elements of intelligence analysis and interpretation. Walsh (2002) offered a first-level decomposition of functionality for Level 2. Powell and Broome (2002) and Powell (2002) indicated the complex set of interrelationships among problems within and between fusion levels characterizing Army intelligence analysis and interpretation suggests a user-centric, holistic approach addressing fusion levels 1 through 5. The present paper characterizes selected key aspects of analysis and interpretation problems and processes based on observations of Army intelligence analysis (in practice) associated with anticipated requirements for the Unit of Action. We analyze the utility and limitations of a computational model of diagnostic reasoning with respect to intelligence analysis and interpretation and identify classes of knowledge that appear to be essential to performing these tasks. The results are considered with respect to their implications for automated support to intelligence analysis and interpretation.