Database Reference
In-Depth Information
Utilising the MISM Model Independent Schema
Management Platform for Query Evaluation
Cornelia Hedeler and Norman W. Paton
School of Computer Science, The University of Manchester
Oxford Road, Manchester M13 9PL, UK
{
chedeler,norm
}
@cs.manchester.ac.uk
Abstract.
Model Management, and its associated operators, provides
generic means for dealing with multiple schemas and the mappings be-
tween them, for example, in the context of multiple heterogeneous data
sources that need to be integrated. One example of a Model Management
framework is the 'Model Independent Schema Management'(MISM) plat-
form. In the context of MISM, algorithms and implementations of various
operators have been proposed that act on a source-model independent
metamodel. However, although the results on MISM indicate how to im-
port and manipulate data from heterogeneous source types, to date no
approach has been proposed to utilise MISM for querying across the mul-
tiple data sources. This paper presents SMql, a query language over the
source-model independent supermodel, presents an algebra into which
the query is translated and presents an approach for rewriting SMql
queries into source-model-specific queries posed over the corresponding
relational or XSD models of the data source to be queried. Thus this pa-
per helps to complete the collection of problems that need to be addressed
to allow source model-independent model management using universal
models in the context of MISM.
Keywords:
Model Management, Query Rewriting, Query Language.
1
Introduction
The vision of model management [5,6] was proposed to address the recurring is-
sues that arise when dealing with data sources, whether multiple heterogeneous
data sources that need to be integrated or a single data source with a continually
evolving schema. In the case of multiple heterogeneous data sources, these could
also be represented using different data models, e.g., relational or XSD. Model
Management aims to provide generic operators that make it easier to manipulate
schemas, that may be associated with, e.g., relational, object-relational or XML
data sources, and the relationships between the schemas. With a view to ob-
taining data model independence, an approach has been proposed to represent
models expressed in various different data models, within the same universal
The work reported in this paper was supported by a grant from the EPSRC.
