This section briefly describes the final architecture proposed for
a Semantic Learning Object repository.
In order to resolve the problems described in the section preceding,
we have proposed several improvements to the SLOR prototype (figure
[17]) :
- First, the SLOR performance is decreased if all inferences related
only to OpenCyc are executed on a semantic web framework. Therefore
to solve this problem, we have incorporated an OpenCyc Server
to hasten general inferences about the common sense reasoning
and prop up the OWL model at the same time (because it’s
just scaffolding). Besides, the SLOR Kernel layer provides a common
set of interfaces to access at low level to the OWL models. Therefore
any change with regard to store and reasoning can be adapted to
different semantic web architectures.
- The second important change is the use of AJAX technology to
enhance the web interaction. Working with AJAX we have developed
smarter pages that achieve good usability levels. From now on,
we will never thing in SLOR without AJAX.
The web services interfaces have the same importance that in the
previous design, several author tools or external agents, if following
the behaviour contract, can access and interact with SLOR. Finally,
we have been working on federation protocols on SOAP to make distributed
queries along some SLOR servers.
Figure 17 - Final proposed SLOR
architecture
|