The Cost of Knowledge Recovery: A Challenge for the Application

Whereas knowledge management relies on processes of knowledge elicitation, there is also a process in which knowledge is "recovered, " typically from archived documents. We conducted a knowledge recovery (KR) effort, going from documents to a structured set of propositions concerning expert knowledge about terrain analysis, discussing landforms, soils, rock types, etc. Assertions and feature associations were recast as over 3,000 propositions. When contrasted with results from previous evaluations of methods of knowledge elicitation, KR was costly in terms of time and effort, suggesting that knowledge-based organizations should make knowledge capture an on-going aspect of work, rather than finding themselves in the "catch-up mode " to recover lost expertise. For both knowledge elicitation and recovery, the knowledge has to be represented in a form that is usable and useful (e.g., instantiation in knowledge bases). We created from the propositions a navigable knowledge model based on over 150 Concept Maps, which were hyperlinked together and to dozens of resources (aerial photos, maps, diagrams, etc.). Such knowledge models are intended to make the "expertise of the past " more useful and usable in training and in performance support