Information Technology Reference
In-Depth Information
user behavior for a more subtle personalization. Their main disadvantage is in the
fact that reuse of the systems is very limited.
The second approach consists of providing a system that is dedicated to
personalization and interacts with a third party for personalization (see e-Person,
BGP-MS, Broadway, NetP 7, Passport.Net and WebSphere). The main functions of
these systems include managing the user profile and the selection of relevant data
that the third party applications are tasked with presenting to the user. These
systems are much more flexible than the previous ones. They are generally used to
personalize several applications likely to be used by the same users. Of course, a
communication protocol is necessary to enable communication between the
personalization system and the third party applications. Besides the advantages that
final users will benefit from (the same profile for different applications, use of user
experiences from one application to another by the system, etc.); the development
costs of personalized applications are dramatically reduced.
The existing systems are generally meant for a particular kind of personalization
and incorporate well-defined methods of personalization. It is very rare to find a
system that ensures personalization of the container as well as the content, for
example. This would require the integration of different (often cumbersome)
methods of collection and management of the user profile in a single system. The
ideal would be to have a personalization that is generic enough (which can support
the different types of personalization) and favors an incremental integration of the
different methods of personalization. As the existing systems do not do this or only
to a small extent, we propose an approach for the design of a personalized
information system (PIS) to satisfy this need
1
.
1 The approach put forward, called PerMet, is based on an ensemble of foundations coming
from state-of-the-art, available in [ANL 06a], several development models (enriched or not
from the point of view of human-machine interactions [KOL 01]) and methods of analysis
and design coming from the literature, representative of several classes of methods. Of these
classes let us in particular note: MERISE (method for the computing study and realization of
business systems) [NAN 01], which is representative of the systemic methods of analysis and
design of information systems;
Two Track Unified Process
[ROQ 07], which is representative
of object-oriented analysis and design methods, such as UP [JAC 99]; WAE (
Web
Application Extension
) [CON 00], which is representative of web applications analysis and
design methods; and AODPU (
Agent-Oriented Design Process with UML
) [CHE 00], which
is representative of methods for the analysis and design of software systems based on
software agents.
