Adaptive Toolkit for Pattern Discovery (ATPD)

We propose to research, design, and demonstrate an Adaptive Toolkit for Pattern Discovery (ATPD). We propose a hybrid approach that combines emerging machine learning techniques with advanced rule based methods to (i) automate the discovery of complex system-of-systems data utilization patterns and (ii) apply the learned usage patterns to dynamically update the reference ontology models being used for enterprise integration applications. The proposed ATPD capability will be used to (a) monitor information use across systems, (b) discover usage patterns, (c) analyze patterns for potential new (and useful) concepts and relations, and (d) offer intelligent assistance to ontology modelers about the integration of newly learned ontology concepts into the evolving reference ontology models. The Phase I effort will (i) establish ATPD requirements, (ii) design the ATPD architecture, and (iii) build and demonstrate a prototype ATPD. The ATPD methods and tools will be incrementally evaluated with test data and application scenarios at the 45th Space Wing, Cape Canaveral Air Force Station (CCAFS). The Phase II effort will harden the ATPD technology and build a focused and field-able ATPD application at 45th Space Wing, CCAFS.