"Knowledge Recovery"


Robert R. Hoffman

David Eccles

Paul Feltovich

Institute for Human and Machine Cognition

University of West Florida




In the development of intelligent systems and also in knowledge management, an important step in Cognitive Task Analysis (CTA) is the initial phase which has come to be called "bootstrapping" (Hoffman, 1987). It is necessary for the cognitive engineer to become familiar with the domain, to establish rapport with the subject-matter experts, and to commence the process of culling information about domain knowledge and domain reasoning. Bootstrapping has been said to primarily involve documentation analysis and initial, informal interviews (Hoffman, et al, 1995). This presentation will discuss an exercise in seeing just how far one can go in the bootstrapping phase toward knowledge preservation in service of the creation of a training/performance aid. It may be quite a bit further than one might suppose.


A study of those features of situations that make decision-making difficult for Army Command and Control (C2) has revealed that coping with terrain is a major source of difficulty. Latent expertise in terrain analysis exists in a number of documents, including two Army field manuals (Department of the Army, 1990, 2002) and a corpus of thousands of propositions concerning terrain, called the "Terrain Analysis Data Base" (TADB) (Hoffman, 1984). Information (from these materials has been recast as Concept Maps. A particular challenge for this project is navigation. ROCK-TA (Representation Of Conceptual Knowledge - Terrain Analysis) will be the largest knowledge model we have made.