Methods for determining object correspondence during system integration

Abstract

Object correlation is a semantic comparison of exported entities from one system to imported entities of another. Current research in search algorithms and artificial intelligence methods for pattern matching can aid integrators in finding these matches. This thesis proposes a two-stage correlation process for resolving various kinds of heterogeneity found in legacy DoD systems to facilitate interoperability. A prototype built using these methods is explained, results compared to current correlation methods, and recommendations made for further improvements. The end of the Cold War and the Defense Reorganization Act of 1986 began a new era of unprecedented cooperation among the U.S. military services and our allies. Increasingly dynamic missions have required warfighters to share information quickly and seamlessly while a decreasing defense budget has left few resources to build the infrastructure needed to implement this information exchange in legacy heterogeneous data systems. One possible solution to achieving interoperability of information systems is Young's Federated Interoperability Model. This model allows system designers to advertise the kinds of information they produce and consume and then automatically provides translation services. Before data and services can be shared, however, integrators must resolve exactly what kinds of data they are providing so that other systems in the network can decide if that data is appropriate for their use. That is the purpose of the proposed correlation algorithm.