Social, Personalized Lifelong Learning - E-Infrastructures and Technologies for Lifelong Learning - page 104

2. authoring for collaboration (adding author
activities, such as defining groups of authors,
subscribing to other authors, etc.);
3. group-based adaptive authoring via group-
based privileges; and
4. social annotation (tagging, rating, and pro-
viding feedback on the content via group-
based privileges).

ditional information to them, as will be detailed
in the definitions below. Also concepts from the
domain model (for instance) can be used by the
environment model, if different content can be
labelled according to the device it is able to work
on, network conditions, etc.
The social component acts vertically, and not
horizontally, as it affects most of the other layers
directly. For example, the resource model layer
includes new entities to describe tags, feedback,
comments, ratings of the actual concepts, and the
relations between these concepts. The domain
model overlays the resource model, and thus inher-
its and manipulates the social activity descriptors.
The goal model includes new entities to describe
the new constraints on the social activities, i.e.
determining who can do what. Moreover, the user
model contains new entities to describe the groups
and the roles (privileges) for these groups, which
will be added to the user model. Additionally, the
adaptation layer holds new entities to handle the
collaborative adaptive strategies. The presenta-
tion layer also contains new entities to describe
how to present information to groups of users.
The adaptation and presentation models use these
elements via data exchange with the package of
socially enhanced models.
Figure 2 also shows the interaction between
the individual models: the social resource, domain
and goal model provide a content-based, metadata-
enriched package to the adaptation model, together
with a social user model, and an environment
model. The adaptation model specifies how the
input from these models is processed, and then
how it is output into the presentation model (what
the learner gets to see) and the update of the user
model (how the information known about a user
is updated).

The Social Reference Model, SLAOS, follows
the multi-layered approach of its predecessors, for
similar reasons: extracting the semantically differ-
ent layers (or models) of a generic system allows
for mapping of different system components onto
the different layers, and thus for a high degree
of reuse of these components, in their interac-
tion with others. For example, a domain model
can be reused with different adaptation models.
These models represent the normalization axes or
principal components of, in our case, a generic
social adaptation system.
SLAOS has taken over the composing models
from LAOS, but refined and extended them, ac-
cording to the social collaborative goals.
Beside the social model, the SLAOS frame-
work encompasses two 'new' models 3 : the Re-
source model and the Environment model.
We have used a resource model, inspired by
the Dexter model (Halasz, 1994), to separate
resources from their domains, and thus allow for
a higher degree of reuse.
Similarly, we have separated the environment
model from the presentation model, to more clearly
separate external factors from what is shown on
the screen. The environment model is subsequently
refined into a physical device model, a network
model and an external environment model, emulat-
ing dimensions from the XAHM model.
The overall structure inherits the conceptual
model based structure from LAOS. The figure
shows also which models are overlaid, such as
the resource, domain, goal, user and environment
models, thus concepts from the resource level
are used in the domain or goal model to add ad-

SLAOS Components:
General Definitions

In the following section, we describe in more for-
mal terms the composition of the SLAOS model,