Using a Semantic MediaWiki to Interact with a
Knowledge-Based Infrastructure
Ian Millard, Afraz Jaffri, Hugh Glaser, Benedicto Rodriguez
School of Electronics and Computer Science
University of Southampton, SO17 1BJ, UK
{icm, aoj04r, hg, br05r}@ecs.soton.ac.uk
ABSTRACT
Facilitating knowledge acquisition is a task that usually re-
quires special purpose interfaces with which users are not
familiar. Providing effective acquisition through a familiar
interface, such as a wiki, can provide a route to acquiring
knowledge for low user investment. We present an archi-
tecture that is being used in the ReSIST project based on
a Semantic MediaWiki integrated with a knowledge base
that allows users to add and view knowledge using normal
Semantic MediaWiki syntax. The architecture aims to facil-
itate the acquisition and representation of knowledge about
resilient systems for users with no experience of knowledge
technologies.
Categories and Subject Descriptors
D.4.3 [Information Systems Applications]: Communi-
cations Applications—Information Browsers
General Terms
Management, Design
1. INTRODUCTION
ReSIST [3] is a Network of Excellence which integrates lead-
ing researchers from the multidisciplinary domains of de-
pendability, security, and human factors. The general fo-
cus of this activity is the advancement and development of
technologies which will ensure that future ubiquitous com-
puting systems have the necessary properties of resilience
and survivability for real world deployment. The project
also aims to create architectures which are tolerant of resid-
ual development and physical faults, interaction mistakes,
and malicious attacks or service disruptions.
In an effort to aid the creation of such systems, the ReSIST
project has embraced the emerging principles of Ontologi-
cal Engineering and the Semantic Web. This has enabled
us to formally describe resilience concepts and the proper-
ties of complex components in detail, as well as informa-
Posters and Demos of the 15th International Conference on Knowledge
Engineering and Knowledge Management, EKAW 2006 Podebrady, Czech
Republic, 2nd-6th October, 2006
tion regarding people, projects and publications. Through
the utilisation of these semantic representations the ReSIST
project is tasked with the creation of a Resilience Knowledge
Base (RKB). The RKB will combine disparate information
sources with suitable user interfaces to provide a central
repository which comprehensively covers all aspects of re-
silient computing and dependable systems. It is envisioned
that the RKB will provide an invaluable resource for both
researchers and students.
The RKB is intended to provide information regarding or-
ganisations that are researching resilient systems; researchers
interested in resilient systems; papers associated with re-
silient systems; faults, errors and failures that have occurred
on IST systems; and other aspects of resilient systems re-
search topics. In addition, knowledge regarding the ReSIST
project itself is recorded, including sub-project activities,
meetings, work package development and management de-
cisions.
However, the task of acquiring semantic information about
an ongoing project from people who are not experts in the
field of knowledge related technologies presents a significant
challenge. Systems must be provided to facilitate as much
incidental knowledge acquisition as possible, while still being
able to gather sufficient knowledge to be meaningful to the
project.
2. SEMANTIC MEDIAWIKI
A significant step in achieving incidental knowledge acqui-
sition has been through the use and customisation of the
newly developed Semantic MediaWiki (SMW) [5]. In ad-
dition to supporting general collaboration between project
members, the SMW provides a means of adding metadata
about the concepts and relations that are contained within
the wiki. This form of ‘tagging’ makes it relatively simple to
turn such annotations into subject, predicate, object triples
that can be stored as RDF and incorporated into the RKB.
Such a system has the advantage of being easy to use for
non experts, but also powerful in the way in which knowl-
edge can be created and stored.
A prototype system has been developed, utilising the SMW
in conjunction with an external 3store [4] RDF repository.
In the SMW, real-world or abstract entities are represented
by an individual page, to which metadata can be added. The
page is therefore represented as the subject resource in RDF
triple form. Relations and attributes are handled differently

Figure 1: ReSIST RKB Architecture
within the SMW, associating the current page with another
SMW resource or a literal value. When a page is saved in
the SMW, custom code is invoked to export the relations
and attributes as RDF and to assert them within the RKB
repository.
For example, a page in the SMW may describe a publica-
tion, to which a project member wishes to be associated as
an author. Utilising the AKT Ontology [1], the following re-
lation may be inserted into the page, specifying the desired
fact.
[[has author::User:Joe Bloggs|Joe Bloggs]]
Entering this special SMW markup within the page will
cause the following triple to be asserted into the RKB:
<http://resist.eu/publications/Bloggs06>
<http://www.aktors.org/ontology/portal#has-author>
<http://resist.eu/people/Joe-Bloggs> .
However, one area in which the SMW lacks good support is
that of namespaces, as general facilities for utilising exter-
nal ontologies, concepts and data-types are yet to be imple-
mented. In a closed world SMW deployment this is not a
problem, and indeed simplifies the input required by users.
Nevertheless, namespaces are vital for disambiguation and
ontological inference, so the export routines apply the rel-
evant namespace prefixes during RDF generation. This is
achieved by using a static mapping between the SMW rep-
resentation of ontological concepts and their external ‘real-
world’ form in the RKB.
The SMW can also be used as a means of exposing knowl-
edge stored within the RKB. For example, pages describing
the classes and properties from external ontologies have been
imported into the SMW, permitting users to view and dis-
cuss the rationale behind each. As well as facilitating col-
laborative ontology development, these representations al-
low users to readily see whether relations and concepts have
been used appropriately when entering semantic markup.
In addition to knowledge obtained through the use of the
SMW, significant efforts are underway to facilitate the ac-
quisition of semantic metadata from external sources. Tools
have been developed to allow non-expert users at each of the
18 ReSIST partner sites to periodically generate RDF data
regarding their institution and its activities. This informa-
tion is then ‘pushed’ to the RKB server through a version
control mechanism and automatically asserted, maintaining
an up-to-date representation of the disparate information
sources. Work has also been done to allow the bulk-import
of information from Cordis, the ACM publications database,
Citeseer and smaller-scale EPrints repositories.
Finally, a generic web-based form interface has been de-
veloped which can be configured to allow the acquisition
of information into a specific ontologically mediated for-
mat. This interface is currently being used to collate user-
submitted data regarding university courses taught to stu-
dents that are related to various aspects of resilient and
dependable systems.
3. FUTURE WORK
The ReSIST project is currently in its ninth month and
the benefits of using the RKB architecture can already be
seen. Content acquisition will be an ongoing process, and
should enable more interesting analysis to be performed once
a more substantial data-set is available. Combined with
this effort is a requirement to develop interfaces with which
novice users can easily explore the RKB, which may poten-
tially include extensions of the work demonstrated in the
CS AKTive Space project [2]. However, the maintenance
of large semantic data sets presents its own challenges, not
least of which are issues regarding referential integrity of
knowledge acquired from multiple sources.
4. ACKNOWLEDGMENTS
This work is supported under the ReSIST Network of Excel-
lence, which is sponsored by the Information Society Tech-
nology (IST) priority in the EU Sixth Framework Programme
(FP6) under contract number IST 4 026764 NOE.
5. REFERENCES
[1] AKT Ontology. http://www.aktors.org/ontology/.
[2] CS AKTive Space. http://cs.aktivespace.org/.
[3] The ReSIST Project. http://www.resist-noe.org/.
[4] S. Harris and N. Gibbins. 3Store: Efficient bulk RDF
storage. In Proceedings of the 1st International
Workshop on Practical and Scalable Semantic Systems,
pages 1–15, 2003.
[5] Semantic MediaWiki.
http://meta.wikimedia.org/wiki/Semantic MediaWiki.