XML schemata inference and evolution

Abstract

This work addresses the automatic generation of conceptual models for XML-oriented databases, which in many cases have little or no support for schemata. Our techniques are based on both an incremental clustering algorithm, which groups together the incoming XML documents according to their structural similarities, and a schema inference method, which maintains dynamically the schema of each detected document cluster. Our proposal takes into consideration the schema evolution. For this purpose, we have adapted the TOODOR document model that describes the temporal properties of the XML document types. © Springer-Verlag Berlin Heidelberg 2003.