Semantics-Preserving Mappings between XML Schemas in P2P Data Integration Systems

Abstract

We discuss the problem of deciding whether there exists a semantics-preserving mapping between two XML schemas or parts of schemas specified by tree patterns. To define semantics of XML data we interpret its schema in an OWL ontology through a semantic annotation (see eg. [1]) and use this annotation to derive complex XML-to-OWL dependencies. There are two steps in our algorithm: (1) schema matching - identification of such maximal parts π1 and π2 of XML schemas, respectively S1 and S2, that π1 is semantically subsumed by π2 (i.e. π1 $\sqsubseteq$ π2); (2) schema mapping - generating a specification describing how data structured according to π1 are to be transformed to data satisfying π2. The existence of matching between subschemas π1 and π2 is a necessary condition for the existence of semantics-preserving mapping between them.