Exemplar-based Latent Semantic Analysis Rational-agent Ontological System (ELROS)

There is a need to automate training, performance evaluations, and knowledge formation for personnel but especially expeditionary forces. The immediacy and high levels of performance required of expeditionary forces can push training tools and techniques to their limits. In response to SBIR AF04-058, 21st Century Systems, Inc. is pleased to leverage our considerable agent-based decision support environment and knowledge representation expertise to rapidly develop an agent-based automated knowledge formation and training system using LSA and rational agent technology. Our system concept is entitled "ELROS", meaning an exemplar-based latent semantic analysis rational-agent ontological system. In our exemplar-based approach to LSA, a populated knowledge repository is used to boost the results of the LSA. Our exemplar-based approach goes beyond concept-based boosting approach by using modal perception categorization and belief statements. ELROS rational agents will provide ready-made training SME's to guide both trainers and trainees through their tasks and amplify their effectiveness by increasing the completion of the highest valuation of tasks first. ELROS rational agents will learn in a passive unsupervised manner, but can also be instructed. Rational agents are proxies for humans and take on their human's needs matrix to build and execute plans within ELROS.