Knowledge Extraction in the SEEK Project Part I: Data Reverse Engineering
J. Hammer, Mark S. Schmalz, Sangeetha Shekar, N. Haldavnekar · 2002
In this report we describe our methodology for knowledge extraction in the SEEK (Scalable Extraction of Enterprise Knowledge) project and highlight the underlying technologies supporting SEEK. In particular, we discuss our use of data reverse engineering and code mining techniques to automatically infer as much as possible the schema and semantics of a legacy information system. We have a fully functional prototype implementation, which we use to illustrate the approach using an example from our construction supply chain testbed. We also provide empirical evidence to the usefulness and accuracy of our methodology.