X-WACoDa - Data Warehousing Design and Advanced Engineering Applications - page 52

REFERENCES On a technical level, SMAIDoC could also be

extended to converse with on-line search engines
(including some semantic Web tools) and exploit
their answers to identify and qualify Web data
sources.
Our choice of an “all-XML” architecture also
leads to address performance problems. Our re-
search in this area shows that using indexes and/
or materialized views significantly improves
response time for typical analytical queries
expressed in XQuery (Mahboubi et al. , 2006;
Mahboubi et al. , 2008; Mahboubi et al. , 2008a).
Further gains in performance can be achieved,
though. For instance, it is widely acknowledged
that indexes and materialized views are mutu-
ally beneficial to each other. We have designed
a method for simultaneously selecting indexes
and materialized views in the relational context,
which we aim at adapting to the XML context.
Finally, our performance optimization strategies
could be better integrated in a host XML-native
DBMS. In particular, the mechanism for rewriting
queries exploiting materialized views would be
more efficient if it was part of the system.
We finally have a couple of perspectives
regarding complex data analysis. First, since the
structure of XML documents carries some relevant
information, we plan to exploit XML structure
mining to, e.g., discover tag relevance. Relevant
tags may then be selected as measures or dimen-
sions in our multidimensional modeling process.
Second, we have underlined with the MiningCubes
software the benefit of associating OLAP and data
mining to enhance on-line analysis power. We are
currently working on extending on-line analysis
with new capabilities such as explanation and
prediction, with the aim of better handling data
complexity.

Baril, X., & Bellahsène, Z. (2003). Designing
and managing an XML warehouse. In XML Data
Management: Native XML and XML-enabled
Database Systems (pp. 455-473). Reading, MA:
Addison Wesley.

BenMessaoud, R., Boussaïd, O., & Loudcher-Ra-
baséda, S. (2006). Efficient multidimensional data
representation based on multiple correspondence
analysis. In Proceedings of the ACM SIGKDD In-
ternational Conference on Knowledge Discovery
and Data Mining (KDD'06) , Philadelphia, USA
(pp. 662-667) New York: ACM Press.

BenMessaoud, R., Boussaïd, O., & Loudcher-
Rabaséda, S. (2007). A multiple correspondence
analysis to organize data cubes. In Vol. 155(1) of
Databases and Information Systems IV - Frontiers
in Artificial Intelligence and Applications (pp.
133-146). Amsterdam: IOS Press.

BenMessaoud, R., Loudcher-Rabaséda, S., &
Boussaïd, O. (2006a).A data mining-based OLAP
aggregation of complex data: Application on
XML DOCUMent. International Journal of Data
Warehousing and Mining , 2 (4), 1-26.

Beyer, K. S., Chamberlin, D. D., Colby, L. S.,
Ozcan, F., Pirahesh, H., & Xu, Y. (2005). Ex-
tending XQuery for analytics. In Proceedings of
the ACM SIGMOD International Conference on
Management of Data (SIGMOD'05), Baltimore,
USA (pp. 503-514). New York: ACM Press.

Beyer, K. S., Cochrane, R., Colby, L. S., Ozcan,
F., & Pirahesh, H. (2004). XQuery for analytics:
Challenges and requirements. In Proceedings
of the First InternationalWorkshop on XQuery
Implementation, Experience and Perspectives
<XIME-P/>, Paris, France (pp. 3-8).