Database Reference
In-Depth Information
Table 1.
Model-generic and model-specific constructs
Metaconstructs
Relational XSD
Abstract
Root Element
Aggregation
Tabl e
StructOfAttributes
ComplexElement
Lexical
Column
Simple Element
Fore ign Key
Fore ign Key Fore ign Key
model, referred to as a supermodel. Utilising the benefits of data model inde-
pendence, it has been suggested recently that dataspace management systems
could utilise model management systems, e.g., for the integration of heteroge-
neous schemas or for dealing with evolving schemas [9]. This, however, requires
model management systems to provide support for query evaluation across the
various heterogeneous data sources.
A prominent example of a source-model independent approach is the 'Model
Independent Schema Management'(MISM) [1,4] framework. In MISM, imple-
mentations of various of the model management operators have been proposed,
e.g., an extended version of ModelGen [3], which in addition to translating
schemas from a representation in one model into an equivalent representation in
another model is also able to translate the corresponding data. This, however,
required the whole database to be loaded, making it only suitable as an off-line
approach. This was addressed for relational data sources in a later proposal [2]
in which the data translation rules presented previously are translated into exe-
cutable SQL statements. The framework was later extended further by proposing
definitions and implementations for Diff and Merge [1] over the supermodel.
However, even though MISM now provides support for managing and inte-
grating schemas represented in various models, it still does not provide sup-
port for querying across the various data sources associated with the integrated
schemas, whereas other model management platforms have been extended to
provide support for query answering (e.g., [17,13]). This paper addresses this
gap by presenting an approach for processing queries in the MISM framework.
To do this we define a query language (
SMql
) and an algebra over the MISM
supermodel and present an approach to evaluating
SMql
queries over relational
and XML sources.
The remainder of the paper is organised as follows. Section 2 introduces the
relevant components of the supermodel of the MISM platform, and Section 3
introduces the query language over the supermodel and the algebra. Section 4
describes the approach for query rewriting. Related work is presented in Section
5 and Section 6 concludes the paper.
2 Background
This section introduces the two levels of schema descriptions of MISM as pro-
posed by Atzeni
et al.
[1,4]. The two levels are the model-specific description,
which contains all the constructs required to represent schemas in a particular
model (see the two UML diagrams in the bottom half of Figure 1 for relational
