NUI Galway – UL Alliance First Annual ENGINEERING AND - ARAN ...

Contextualized Application Reasoning Environment (CARE)
Ratnesh Sahay, Antoine Zimmermann, Ronan Fox, Axel Polleres and Manfred Hauswirth
E-mail: firstname.lastname@deri.org
Abstract
The article briefly describes a Contextualized Application
Reasoning Environment (CARE) system. We discuss the need
of such a reasoning system for Health Care applications. We
highlight core research issues of a knowledge-based system
in dealing with various types of knowledge (e.g., profile,
policy, temporal, spatial) as well as knowledge that is contextdependent.
Finally, we discuss the current status and future
work in developing the CARE system.
1. INTRODUCTION
The notion of context has a very long history within
several research communities, leading to vast studies about
how to define a context, how to take into account information
coming from the context, how to contextualize knowledge
and local information, etc. On the other hand, information
systems that are deployed across organisations are usually
context-dependent, for instance, constraints (e.g., profile, policy,
preference, temporal, spatial) expressed by particular
organisations are usually applicable within the organisations
and their appropriate interpretation is meaningful only when
contextual knowledge (e.g., place, event, time) is taken into
account. In this article, we discuss the issue of context and
constraints from two perspectives (1) Web reasoning and (2)
when reasoning is applied on concrete domain like Health Care
and Life Sciences (HCLS).
2. CONTEXT AND CONSTRAINTS: WEB REASONING
Semantic Web and the reasoning mechanism behind ontological
knowledge bases are centralised in the way data
and schemas are accumulated and processed. Ontologies—the
pillars of semantic Web framework—are good in describing
general invariant concepts and mappings or relations among
those concepts. However, when the semantic Web enabled
applications use data and schema that are distributed, heterogeneous
and multi-contextual, then interoperability between
interacting applications are effected adversely[1].
3. CONTEXT AND CONSTRAINTS: HEALTH CARE
APPLICATIONS
HCLS has been one of the primary field of application for
knowledge representation and reasoning systems. In the past
researchers have tried to formalise and integrate the knowledge
bases in HCLS systems and many of the successful systems
in earlier times were centralised and limited to sub domain
or particular application of a HCLS domain [2]. Nowadays,
HCLS has become more global and distributed in terms of
its use by related stakeholders[3]. Therefore, interoperability
enabler, i.e., Web reasoning systems, need to be extended in
a manner that they can reason various consequences while
aggregating and interpreting global knowledge in conjunction
with local information or constraints.
4. CARE
We are developing the CARE system that can identify the
context of knowledge bases with a mechanism for handling
constraints that do not lead to undesired interactions. This
approach is partly inspired by [4], where axioms are separated
into two T-Boxes, one for ontological axioms, the other for
integrity constraints. This way, we define a local T-Box as a
pair 〈D, P 〉, where D describes a domain ontology, and P
represents the internal policies. If several local ontologies and
policies exist, the overall knowledge is a distributed system
(〈Di, Pi〉).
In CARE domain ontology (Di)
can import other domain
ontologies (i.e., import closure Di Dj). Internal ontologies
(Pi) are context-dependent (i.e., constraints) and allowed only
to refer domain ontologies (Di) of their context. While reasoning,
especially for
local entailments, import closure of
domain ontologies (Di Dj) and internal ontology (Pi) of the
requesting system will be used. CARE allows global and local
(context-specific) entailments over distritbuted knowledgebases.
46
5. CONCLUSION
We have described use of CARE for health care applications.
The initial prototype of CARE is able to categorise
between differen types of knowledge (e.g., general, policy)
and use contextual information for avoiding undesired inconsistencies.
The future task is to include modular semantics for
the CARE system.
6. ACKNOWLEDGEMENT
This work has been funded by Science Foundation Ireland,
Grant No. SFI/08/CE/I1380 (Líon-2).
REFERENCES
[1] T. Berners-Lee and L. Kagal, “The Fractal Nature of the Semantic Web,”
AI Magazine, vol. 29, no. 3, pp. 29–35, 2008.
[2] P. Szolovits, Artificial Intelligence in Medicine. Westview Press, Inc.,
1982.
[3] A. L. Rector, R. Qamar, and T. Marley, “Binding ontologies and coding
systems to electronic health records and messages,” Applied Ontologies,
vol. 4, no. 1, pp. 51–69, 2009.
[4] B. Motik, I. Horrocks, and U. Sattler, “Bridging the gap between OWL
and relational databases,” in Proc. of 16th International Conference on
World Wide Web, WWW 2007. ACM Press, 2007, pp. 807–816.