From: Lucas Kreger-Stickles (lucasks@cs.washington.edu)
Date: Mon Apr 19 2004 - 11:42:20 PDT
*Review:*
*Relational Databases for Querying XML Documents:*
* Limitations and Opportunities
*
In their paper "Relational Databases for Querying XML
Documents:Limitations and Opportunities" the authors present techniques
for converting semi-structed XML data with coresponding DDTs into
Relational Database tables. In addition, they present techniques for
converting queries on that data into SQL queries and for converting the
results back into XML style data.
The authors argue that such a technique may have merit given the quickly
growing promence of XML (which indicates that people will want to use
and querry data formatted in such a way) and the over 20 year research
and development devoted to relational databases.
While the authors succed at converting and handeling the vast majority
of XML queries via a relational database system, they conclude that such
a process has many potential problems. In general they point to the
fact that many relativly simple XML querries turn into large or
multipart SQL queries that involve a great many joins between them.
However, the authors also point out that just because such querries are
expensive in a relational setting does not garantee that such queries
would be easier to execute in an XML setting.
Finally, the authors suggest possible extensions to relational databases
that would make such conversions more straight-forward and querries less
costly. They suggest: Support for sets, untyped references, information
retrieval style indecies, flexible comparison operators, multiple-querry
optimization, and more powerfull recursion.
In the end the authors indicate that it is an open question as to
whether it is better to try and map XML to relational data (and thus use
relational techniques learned over sevral decades) or to map the
teechniques we have learned regarding relational data to XML querries as
best we can.
Overal I thought it was a well thought and well argued paper. Far too
often it appears that reserachers are so determined to present their
ideas as effective that much valuable insight gets lost in the effort to
insist that a particular technique is worhtwhile. Here the authors
manage to succeed at mapping XML into a relational database and then
have a well formulated discussion as to the merits and shortcommings of
such an approach.
This archive was generated by hypermail 2.1.6 : Mon Apr 19 2004 - 11:42:25 PDT