Database Reference
In-Depth Information
4 Related Works
Most works that discuss the transformation of XML documents via XQuery focus on
an efficient evaluation that can handle uncompressed XML only. A first example is
[8][8] that discusses the unnesting of FLWOR expressions, while [9] introduces an
algebra for XQuery and shows, how unnesting can be performed within this algebra.
Further algebras for XQuery were introduced by [10] and [11]. The work of
Fernández and Siméon was realized in an XQuery processor called Galax [12]. With-
in Galax, the internal data model is implemented for three different kinds of XML
representations, but none of these is a compressed XML representation.
The evaluation of XQuery instructions on top of compressed XML data is
performed by the approaches XQueC [2] and Bisimulation [3]. A document that is
compressed via Bisimulation can be transformed with the help of XQuery into a com-
pressed target document. In this approach, the XQuery language is restricted - similar
as in our approach - to for, let and return expressions. In contrast to our approach,
Bisimulation [3] supports XPath expressions consisting of child axis steps only, whe-
reas our approach supports all forward axes except the following axis.
XQueC [2] proposes an XML representation that is optimized for an efficient
transformation via XQuery. The structure compression as well as the data compres-
sion is chosen in such a way that path queries can be evaluated efficiently on it, but in
return, the compression ratio reached by XQueC [2] is not as strong as by other com-
pressors. The approaches being used in XQueC [2] appear to be non-applicable to
other compression techniques.
Our approach contains parts that are generic, while other parts are specific for the
compressed XML representation being used. Whenever another compressed XML
representation supports the specific parts, our approach to transforming compressed
XML representations can be applied to the other compressed XML representation as
well.
To summarize, to the best of our knowledge, our approach is the only approach
that provides a generic XQuery transformation of compressed XML data which is
applicable to all compressed XML data formats that support the basic navigation steps
first-child, next-sibling, end-of-parent, self::label, and that support inserting nodes as
first-child or as next-sibling, and that optionally support a copy operation for com-
pressed data.
5 Summary and Conclusions
Whenever the data volume of XML files to be exchanged or stored and to be trans-
formed is a bottleneck of an application, transforming compressed XML is a promis-
ing alternative to decompressing compressed data back to XML prior to transforming
it, and eventually recompressing the transformed XML data afterwards.
In this paper, we have presented a generic XML transformation approach that al-
lows evaluating queries belonging to a subset of XQuery on all the compressed or
uncompressed XML representations that support at least a small set of basic opera-
tions like the navigation via the binary XML axes first-child, next-sibling and end-of-
parent, and like the insertion of nodes.
