Operationally significant patterns of association

An association between two persons is an observation or reported fact that can reasonably be assumed to indicate they know of, and probably communicate with, each other. Much of the information on known/suspected terrorists comprises collections of such associations absent information that would support determination of whether any particular set of associations evinces cooperative involvement in pursuit of a common goal or objective. A major challenge for information technology is to develop tools for discovering such stronger relationships in large collections of associations gathered from disparate sources. To this end, this paper describes four patterns of associations that are indicators of possible organizational activity, regardless of the means of their discovery. It describes for each: i) techniques for visualizing the pattern in the data; ii) efficient means of discovery of instances of the pattern in very large link data bases; and iii) procedures for validating instances of the pattern as being attributable to organizational behavior rather than a chance encounters. It further describes results of analyses of real-world data that demonstrate the operational utility of inferences derived from the patterns.