A General Framework for Inclusive Lifelong Learning in Higher Education Institutions with Adaptive Web-Based Services that Support Standards - E-Infrastructures and Technologies for Lifelong Learning - page 40

IMS-LD) and (ii) learning management services
(which require an extension of the IMS-LD to
describe them). Due to the operational nature of
the services, they can be defined as adaptive work-
flows described in terms of IMS-LD guidelines
that implement psycho-educational strategies to
support learners who encounter barriers in their
interaction with HE institutional services. The goal
here is to adapt these services to the users' needs
in an inclusive way through the full life cycle of
service adaptation.
However, IMS-LD presents limitations, es-
pecially for the accommodation of the learning
management services. Even for the traditional
learning services (despite IMS-LD claims to
support all pedagogical scenarios) practice has
proven that there are restrictions. First, the learn-
ing resources and services, which are needed to
complete the design, are not modelled in detail
within the specification but are pointed to exter-
nal spaces (Boyle, 2009). Second, the condition
model of IMS-LD presents scalability issues with
implication on maintainability of units of learning
(Gutierrez et al., 2008). Third, there exist other
practical considerations regarding the excessive
size of the resulting files when a rich sequencing
strategy is authored (Gutierrez et al., 2008). Fourth,
it does not properly support cycling workflows
(i.e. reflexive cycles), where the course author
would like to make the learners go back to some
activity already performed (Gutierrez et al., 2008).
Fifth, there is not full support for collaborative
activities, since there is no way to model groups of
users doing the same activity. Some approaches to
cope with some of these limitations are described
in de la Fuente et al. (2007). Moreover, there
are limitations related to the accessibility of the
players and the specification itself which do not
consider in advance accessibility issues. To cope
with the latter our research group is participat-
ing in both the accessibility improvement of an
IMS-LD player and the combined management
of the specification with other user-centered and

accessibility focused specifications and standards,
such as those that manage user profile informa-
tion (see below).
In this area, the work of our research group is
proving that specifications such as IMS Learn-
ing Design can contribute to the derivation of
guidelines for the delivery of some HE services
for people with disability taking into consider-
ation psycho-educational requirements (e.g. in
the enrolment process). Moreover, it is making
contributions for the implementation of appropri-
ate tools to support these users (e.g. for the acces-
sibility of the player interface and the integration
with the information stored in the user profile).
However, universal design approaches do not
suffice (Martinez-Normand, 2007) as in practical
situations it is not possible (in terms of time, ef-
fort and knowledge) to specify the whole design
in advance. Moreover, it is not feasible either to
collect directly from the users the values of the
required attributes in order to build the model
that the adaptation tasks will consider. There is
a need for intelligent support that would analyze
the learners' interactions at run time and process
them to learn their usage preferences in order to
provide runtime adaptation. In particular, artificial
intelligence techniques, such as data mining to
extract the knowledge from the interactions, and
machine learning to classify users according to
their behaviour, can be used when dynamic sup-
port is required. In turn, collaborative filtering
techniques are useful to offer recommendations
to the users based on other users' experience and
the similarity of user models (and thus, needs and
preferences) among the users.
Evaluations with users justify the need of the
dynamic support at runtime (Santos and Boticario,
2008a) when trying to support inclusive scenarios.
Our research group has proposed the integration of
recommender technology for providing dynamic
adaptations that overcomes the limitations of the
universal design approach. To offer these recom-
mendations in inclusive adaptive standards-based