X-WACoDa - Data Warehousing Design and Advanced Engineering Applications - page 40

INTRODUCTION (Bray et al. , 2006) bears many interesting features

for representing complex data (Boussaïd et al. ,
2007; Boussaïd et al. , 2008; Darmont et al. , 2003;
Darmont et al. , 2005). First, it allows embed-
ding data and their schema, either implicitly, or
explicitly through schema definition. This type
of metadata representation suits data warehouses
very well. Furthermore, we can benefit from the
semi-structured data model's flexibility, extensi-
bility and richness. XML document storage may
be achieved either in relational, XML-compatible
Database Management Systems (DBMSs) or in
XML-native DBMSs. Finally, XML query lan-
guages such as XQuery (Boag et al. , 2007) help
formulate analytical queries that would be difficult
to express in a relational system (Beyer et al.
2004; Beyer et al. , 2005). In consequence, there
has been a clear trend toward XML warehousing
for a couple of years (Baril & Bellahsène, 2003;
Hümmer et al. , 2003; Nassis et al. , 2005; Park et
al. , 2005; Pkorný, 2002; Vrdoljak et al. , 2003;
Zhang et al. ,2005).
Our own motivation is to handle complex data
into a complete decision-support process, which
requires their integration and representation under
a form processable by on-line analysis and/or data
mining techniques (Darmont et al. , 2003). We have
already proposed a full, generic data warehousing
and on-line analysis process that includes two
broad axes (Boussaïd et al. , 2008):

Data warehouses form the basis of decision-
support systems (DSSs). They help integrate
production data and support On-Line Analyti-
cal Processing (OLAP) or data mining. These
technologies are nowadays mature. However, in
most cases, the studied activity is materialized by
numeric and symbolic data, whereas data exploited
in decision processes are more and more diverse
and heterogeneous. The development of the Web
and the proliferation of multimedia documents
have indeed greatly contributed to the emergence
of data that can:

• Be represented in various formats (data-

bases, texts, images, sounds, videos...);
Be diversely structured (relational databas- •

es, XML documents...);
Originate from several different sources; •

• Be described through several channels or

points of view (a video and a text that de-
scribe the same meteorological phenom-
enon, data expressed in different scales or
languages...);
Change in terms of definition or value •

over time (temporal databases, periodical
surveys...).

We term data that fall in several of the above
categories complex data (Darmont et al. , 2005).
For example, analyzing medical data regarding
high-level athletes has lead us to jointly exploit
information under various forms: patient records
(classical database), medical history (text),
radiographies and echographies (multimedia
documents), physician diagnoses (texts or audio
recordings), etc. (Darmont & Olivier, 2006; Dar-
mont & Olivier, 2008)
Managing such data involves lots of differ-
ent issues regarding their structure, storage and
processing (Darmont & Boussaïd, 2006); and
classical data warehouse architectures must be
reconsidered to handle them. The XML language

• Data warehousing, including complex data

integration and modeling;
Complex data analysis. •

More precisely, the approach we propose con-
sists in representing complex data as XML docu-
ments. Then, we recommend an additional layer to
prepare them for analysis. Complex data under the
form of XML documents are thus multidimension-
ally modeled to obtain an XML data warehouse.
Finally, complex data analysis can take place from
this warehouse, with on-line analysis, data mining
or a combination of the two approaches.