Modeling Glocal Search in a Decision-Making Process.

This work focuses on the formalization of medical practices in chronic inflammatory bowel disease diagnosis as a contextual graph to identify a consensual methodology. Expert knowledge is more than domain knowledge because expert knowledge emerges from a contextualization process and expertise appears as “chunks of contextual knowledge”. The knowledge acquisition phase and the modeling of the expertise were organized around three workshops about the choice of diseases to study, the observation of how pathologists worked, the identification of similarity of pathologists’ approaches and the proposal of a unified view. The “knowledge manager” was a physician knowing the expert knowledge used by experts but without their experience. A glocal search was identified in the decision-making process with a global exploration for detecting zones of interest, and a zoom inside zones of interest. This search concerns contextual elements at different levels of granularity as identified from the analysis of digital slides. The rest of the decision-making process includes the application of a set of criteria that are managed by a voting system that offers flexibility to address the variety of expert approaches. We discuss the role of the glocal approach in other domains like Control & Command rooms for the army and subway-line monitoring.