Latent Semantic Analysis: A technique for enhancing use of OA data in evaluating occupational restructuring

Darrell Laham, Winston Bennett, Jr., and Thomas K. Landauer

Abstract

Military organizations are increasingly faced with rapid changes in technology and missions, and need constantly changing mixes of competencies and skill. Assembling personnel with the right knowledge and experience for a task is especially difficult when there are few experts, unfamiliar devices, redefined goals, and short lead times for training and deployment. Large civilian organizations face similar challenges in adapting to international competition, new technologies, and organizational re-alignments. When too few adequately trained personnel are available for suddenly critical tasks, organizations need the ability (a) to identify existing personnel who could perform the task with the least training, and (b) to create new training courses quickly by assembling components of old ones. New LSA-based agent software helps to identify required job knowledge, determine which members of the workforce have that knowledge, pinpoint needed retraining content, and maximize training and retraining efficiency. The LSA-based technology extracts, represents and matches information about people, occupations, and experience contained in textual databases. To demonstrate and evaluate the system, we analyzed the tasks and personnel in three Air Force occupations. We measured the match of each airman to each task and estimated how well each airman could replace another. We also demonstrated the potential to match knowledge subcomponents needed for new systems with ones contained in training materials and those possessed by individual airmen. Our research provides results that demonstrate that LSA can successfully characterize tasks, occupations and personnel and measure the overlap in content between instructional courses covering the full range of tasks performed in many different occupations. This research shows the potential for LSA-based methods to identify ways in which occupations might be reorganized to increase training efficiency, improve division of labor efficiencies, or redefine specialties to produce personnel capable of a wider set of tasks and easier reassignment. The natural language query design intrinsic to LSA eliminates the known problems inherent in keyword matching of field-restricted databases.

Full Paper (HTML)