Ontology-Based Database Approach for Handling Preferences - Data Warehousing Design and Advanced Engineering Applications - page 253

Table 1. Preference definition approaches in Databases, Semantic Web and and Data Warehouses do-
mains

Preferences
Model Level

Storage of Preferences
Model

Author Domain (DB/SW/DW) Approach Query Operator

Kießling
(2002-2003)

DB Qualitative Logical Preference SQL NO

Chomicki
(2003)

DB Qualitative Logical Winnow NO

Agrawal &
Wimmers
(2000)

DB Quantitative Logical NO NO

Koutrika &
Ioannidis
(2006)

DB Quantitative Logical NO NO

Das et al.
(2006)

DB Quantitative Logical NO NO

Siberski et al.
(2006)

Boolean/ Scoring
preferences

SPARQL
Clause Preferring

SW Semantic NO

Sieg et al.
(2007)

Ontological User
Profiles

SW Semantic NO NO

Gurský et al.
(2008)

Fuzzy based/
Ontology

SW Semantic NO NO

Toninelli et al.
(2008)

Middleware-level/
Meta Model

SW Semantic NO NO

Bellatreche
et al. 2005
Mouloudi et
al. 2006

DW Qualitative Logical NO NO

mantic Web, except the works done by (Bellatreche
et al., 2005; Mouloudi et al. 2006). They present
a personalization framework for OLAP queries.
They give end user the possibility to specify her/
his preferences (e.g., the presence of a given di-
mension of a data warehouse in the final result)
and her/his visualisation constraint to display
the result of an OLAP query. The visualisation
constraint represents the size of device (PDA,
mobile phone, etc.) used to display the result of
a query. The authors present some issues on the
impact of preferences on physical design of a data
warehouse (data partitioning, index and material-
ized view selection). This work did not present
query operator handling preferences.
The presented approaches are summarized in
Table 1. We classified them according to 5 cri-
terions. The first criterion indicates whether the

considered approach is presented in the context
of Databases (DB), Semantic Web (SW) or Data
Warehouse (DW). The second criterion indicates
the followed approach (qualitative, quantitative,
etc.). The third criterion indicates at what level
(physical, logical, semantic) preferences are de-
fined. The fourth criterion indicates whether there
is a specific operator for querying with preferences.
Finally, the last criterion indicates whether the
considered approach offers the possibility to store
physically the preferences model.
By analysing the above table, we note that
the notion of preferences as considered in the
databases domain is introduced mainly at the
logical level. Preferences are defined on top of
the manipulated models themselves and conse-
quently, they depend on the model they extend.
In the Semantic Web domain, even if an ontol-