Towards Achieving Semantic Interoperability in eHealth Services - E-Healthcare Systems and Wireless Communications - page 393

one for all or incorporating it in the functional
interface, thus over-complicating or limiting its
extensibility respectively. Instead, this approach
makes the architecture potentially re-usable across
other domains that require such requirements.
A domain-based service-oriented architecture
for interoperability in eHealth presents additional
unique challenges. For instance, data sources can
be large and complex, served by autonomous,
slowly evolving legacy systems. Data sources
reside on different and varied platforms and their
records can potentially range from several thou-
sands to hundreds of thousands, if not millions.
Many have low-bandwidth access and all are used
to directly support the healthcare process and
potentially already overloaded and thus access-
ing them in real-time by other services needs to
be taken under strict conditions. This effectively
limits available processing capabilities and offers
differing reliability when compared to dedicated
function-focused Internet web services. However,
for the purposes of this paper, we will be focus-
ing on interoperable semantic composition in the
eHealth domain. In particular, we aim to address
service interoperability and the pressures placed
on the service engineering process through the
use of semantic representation in data-intensive
services in complex, sensitive and closed domains,
focusing on eHealth. To do so, we propose that the
service-oriented approach must be supplemented
by a (run-time) computationally interpretable
domain-based semantic representation, for both
the data and its access constraints.

kiraju, R. et al (2005)]. OWL-S is based on the
earlier work on DAML-S, which uses ontologies
to define the requirements and capabilities of web
services that can be reasoned about at run-time
to dynamically discover appropriate services and
bind to them. WSDL-S proposes an extension to
the widely used WSDL standard to enable semantic
expressivity. It defines a mechanism to associate
semantic annotations with web services that are
described using WSDL. WSDL-S assumes that
service semantics are described in formal models,
referred to as semantic models , which have link
annotations into WSDL documents. However,
both OWL-S and WSDL-S are bound to WSDL
and only include semantic information that de-
scribes the service specific elements, such as the
definition of preconditions, inputs, outputs and
operations. These definitions are oriented towards
function-intensive services and lack the needed
expressivity richness needed for data-intensive
services, especially in complex domains, where the
focus is on defining or representing the data itself.
These approaches' focus has been on creating
autonomous, self-describing web services that ex-
ist independently in the market place, more suited
to application-oriented or function intensive web
services that provide concrete services to other
consumers. However, for data-intensive web
services that cover specific domain needs, these
approaches can be expensive and less suitable.
For data-intensive web services, the main effort in
creating semantic interoperability between them
will be focused on, describing the heterogeneity,
access policies etc. of the underlying data itself
in computer-interpretable formats, rather than the
functions of the web services themselves. Within
well-defined and potentially closed and sensitive
domains, such as health, the actual importance
and the value is associated with the data itself
and the mechanisms of its access. But also these
required semantic descriptions are often more
comprehensible and compatible within the domain
and thus the need is to find a domain-dependent
mechanism to facilitate their interoperability and
management.

SEMANTIC SERVICE INTEGRATION

Several approaches have been proposed to enable
dynamic composition of web services based on
the semantic expression of their functionality
[Ankolekar, A. et al (2002), Martin, D. et al
(2007), Martin, D. et al (2004), Akkiraju, R. et
al (2005)]. The more recent are the efforts from
the semantic web community [Martin, D. et al
(2007), Martin, D. et al (2004)] and W3C [Ak-