Assembling people with the right knowledge and experience for a mission is especially difficult when there are few experts, unfamiliar devices, redefined missions, and short lead times for training and deployment. This effort is developing a prototype intelligent agent that helps to identify required mission knowledge, determine if current AF personnel have that knowledge, pinpoint needed retraining content to offset knowledge deficits, and maximize training and retraining efficiency. The research objective is to develop, scale up and test methods based on Latent Semantic Analysis (LSA), a new knowledge representation technology whose relevant capabilities were demonstrated in Phase I. LSA is a machine-learning method for authomatically extracting knowledge from existing information about people, occupations, and experience. The Phase I prrof-of-concept pre-prototype analyzed tasks in three AF occupations, measured the match of each airman to each task, estimated how well each airman could replace another, and illustrated the potential ability to identify and match those knowledge subcomponents needed for new systems, those contained in training materials, and those possessed by individual airmen. Phase II will scale up to practical coverage, and expand and extend the applications to airman identification methods and creation of training materials, and to a wide range of occupations and tasks.
Keywords: DATA MINING INTELLIGENT AGENT KNOWLEDGE REPRESENTATION LSA LATENT SEMANTIC ANALYSIS MACHINE-LEARNING
