Humans possess a highly developed ability to reason directly with diagrammatic information. Endowing a machine with similar capabilities could prove beneficial in the development of an artificially intelligent agent that interacts with a real world environment, rife with diagrammatic information, with a higher degree of autonomy than those without such capabilities. The goal of this research is to develop a theoretical basis for computing directly with diagrammatic representations (although the PI recognizes that diagrammatic representations may need to be augmented by other representations to completely represent many problem domains). More specifically, this research proposes to examine how the spatial and temporal coherence often exhibited by groups of related diagrams could be leveraged for computational purposes. Using concepts gleaned from set theory, image processing theory, color theory, and the theory of computation, like diagrams are combined in ways that produce new like diagrams that infer information implicit in the original diagrams. This approach has been successful in producing diagrammatic solutions to problems in diverse contexts including heuristic development, spatial configuration, inductive learning, scheduling, and DNA sequencing. The PI and his students will consolidate this previous work with other research concerned with diagrams and computations, expand the current theory through rigorous investigation of its relationship with image processing theory, and test the applicability of the theory through the design and implementation of a diagram information system that allows users to pose queries concerning diagrams in a specified domain whose responses require information inferred from diagrams. http://morpheus.hartford.edu/~anderson/personal
