Two main evidences motiveted the project: the forecast of an exponential increase of
adaptive hypermedia systems and the difficulty of identifying the user's relevant conceptual dimensions, that are, indeed,
the keys for defining reliable and predictive user models in these systems.
|
 |
The idea of the project is that these conceptual dimensions are closely related with the goals and sub-goals of the
adaptive systems, with the tasks performed to obtain such goals and with the features of the application domains.
Moving from that, and adopting a bottom-up approach, the project aims at creating a taxonomy of goals and features
for different application domains with the final goal of introducing an ontology for conceptual dimensions in user modeling.
The benefit expected from the framework is that of supporting designers in two ways:
-
Creating some guidelines (definition of a method) to simplify the analysis required when designing new adaptive applications, i.e. guidelines that can suggest the relevant user dimensions, given the goals and the domain features of the application.
-
Building user modeling components that describe the user modeling dimensions, in the form of stereotypes, and that can be reused across different applications.
The ontology has in fact to become a library of modular knowledge sources. The choice of representing each dimension as a
set of stereotypes expressed in a standard formalism is oriented in this direction, and is due to the need of using a declarative,
neutral and easy to understand language. Each stereotype is formed by a set of slots, corresponding to features of the user
modeling, and each slot contains a set of pairs, corresponding to:
-
the linguistic value of the feature;
-
the probability of the linguistic value of the feature, given the stereotype, namely the frequency of
this feature in the population belonging to this one stereotype.
THE FRAMEWORK EXTENT
The whole project is founded on the principle that stereotypes are extremely useful in several situations:
-
for the initialisation of the user models
-
when poor information is available about users
-
in applications with no session memory (es. kiosks, showrooms, etc.)
Anyway, the objective of the framework is to be useful not only for the library of stereotypes, that can be re-used
in several applications, smoothing moreover the communication and the share of knowledge with other systems, but
especially for the provision of a method and a predefined analysis for identifying the dimensions relevant for user
modeling.
For more information about this project, please contact
Ilaria Torre

|