Database Reference
In-Depth Information
condition (Koutrika and Ioannidis, 2004).Atomic
preferences can be combined and used to derive
implicit preferences
. For example, considering the
same table Hotel, the expression <
Hotel
.
name
=0
Sophitel,
0.8 > indicates that the interest degree
of
Sophitel Hotels
is 0.8.
Following the quantitative approach as well,
Das et al. (Das et al., 2006) propose a presentation
based preferences of user profiling. These prefer-
ences define an order relation between the tuples
returned by a given query. In this case two tuples
are compared after accessing data sources. The
comparison is based on a set of selected attributes.
The presentation is expressed by the selection of
a set of relevant attributes and by displaying the
corresponding tuples in the tables and ranked
according to their importance.
These approaches are strongly linked to the
logical model of the database. Therefore, a good
knowledge of this logical model is required for
an efficient exploitation of these models. They
implement various operators including
winnow
(Chomicki, 2003), Top(k) (Theobald et al., 2004),
Skyline (Brzsnyi et al., 2001), Pareto (Viappiani
et al., 2006) and Preferring preference operator
(Kießling, 2002) which are applied on the database
table's columns.
Different models of preferences have been
proposed in the literature (Siberski et al., 2006;
Gurský et al, 2008; Toninelli et al., 2008; Sieg
et al., 2007).
The
Local Preference Model
is proposed by
Gurský et al. in order to model complex user pref-
erences (Gurský et al, 2008). They consider that
complex preferences reflect more accurately real
life preferences. They use a fuzzy based approach
for preferences description. Firstly, nominal and
ordinal attributes are used to define local prefer-
ences. Then, their combination with user's local
preferences produces global preferences. For
example, the global preference
good hotel(x)
can
be defined by the combination of the two local
preferences expression
good price(x)
and
good
starRating(x).
Toninelli et al. introduce the
Ontology Based
Preference Model
approach by defining a meta
model (Toninelli et al., 2008). In this approach,
value and priority preferences are specified. For
example, to find high standard hotels, the quality
of the service must be a priority.
Siberski et al. propose an extension of the
SPARQL query language (SPARQL, 2008) by
introducing the
Preferring
modifier (Siberski et
al. 2006). This modifier allows users to define
Boolean and scoring preferences. For instance,
the expression
< rating = excellent >
specifies
an excellent hotel.
Sieg et al. present an ontology based approach
for personalising Web information access (Sieg
et al., 2007). The user interests are captured im-
plicitly by a context defined through the notion of
ontological user profiles. This context model for
a user is represented as an instance of reference
domain ontology. The concepts of the ontology are
annotated by
interest scores
derived and updated
implicitly according to the user's behavior.
Preferences in Semantic Web
The notion of preference is a crucial issue in
the Semantic Web area. The Semantic Web is
a vision where data on the Web can be defined
and linked in such a way that they can be un-
derstandable and be processed by machines
and not only by humans (Berners-Lee et al.,
2001). The Internet enables user to access a vast
amount of available information, but this often
results in the inability to find the relevant and
needed information. Thus, in order to provide
more accurate information, the various Internet
portals, and more generally the Semantic Web
must support profile based search and prefer-
ences based browsing.
Preferences in Data Warehouses
There is a few works on preferences in the data
warehouse context compare to database and Se-
Search WWH ::
Custom Search