![](bg4.png)
property these objects must exist in order to be able
to be referenced by object properties. This is not
trivial which is reflected by the fact that various
tools explored and described in the previous section
do not provide this feature at all.
In order to solve this problem our mapping in
IwOnto is split into two phases. In the first phase all
needed objects for an object property should be
generated. Because of possible dependencies this
again is not trivial. If class A has an object property
which range is class B, this implies a dependency of
class A on class B. When building the dependency
graph there can be chains and circles in it. To handle
circles at least one link must be broken by removing
one object property in the circle. This results in the
loss of information that should be included in a later
stage.
In the second phase the generated objects can be
used. This is also not as easy as it might sound. One
needs to identify the generated objects of the first
phase clearly and without ambiguity based on the
information provided by the intermediate table of the
second phase. We currently have different
approaches we implemented and tested to do so: a)
based on the name of the object that needs to match
the mapped attribute value, b) by a special data
property as identifier for it, or c) a more fussy
solution scanning all data properties in order to find
a match. Currently we favour the first approach
based on the name of the object because it is the
most secure approach and this way we do not
generate an IwOnto specific solution. However,
none of these approaches is perfect, and we are also
investigating additional and more flexible ways, e.g.
based on constraints, to map from the attribute
values of the intermediate table to objects already
generated.
4 CONCLUSIONS AND FUTURE
WORK
Existing mapping and export tools for migrating
relational data to an existing ontology did not match
our needs. We therefore implemented IwOnto. There
are two main differences compared to existing tools
a) it is based only on SQL and does not enforce
users to adapt additional languages like SPARQL
and b) the ability to move some of the mapping
problems into intermediate SQL statements rather
into the mapping itself, helps the user to split his
problems into manageable portions.
It is already possible to handle object properties
with IwOnto, which was problematic with many
other existing tools. However, generating the
according objects in case of longer chains is a time
consuming task and does not fit our idea of an easy
to use tool. Solving this in a sufficient and user
friendly way is currently our main challenge.
The second main focus we have is the good
usability of IwOnto. In the latest implementation we
therefore improved the way how SQL statements
can be entered. Now a statement can be tested and
the user can see immediately the result of his
statement in order to be sure he generated the correct
intermediate table.
Currently we use IwOnto in the domain of
archaeology and the size of relational databases we
need to handle at the moment consists of just a few
hundred findings. Therefore, performance issues are
not our main focus. However, IwOnto is not bound
to this domain and once we solved our current
issues, we will have a closer look on performance. A
first prototype of IwOnto is available under GPL 3.0
at SourceForge.
ACKNOWLEDGEMENTS
The authors would like to thank Steffen Försch,
Marco Bender and Alexander Rezun for their
contributions to IwOnto.
REFERENCES
Bachmann, M. (2011). Informationsintegration mit
Ontologien am Beispiel von archäologischen Daten,
Diploma Thesis at Goethe University Frankfurt am
Main.
Barrasa, J., Corcho, Ó and, Gómez-Pérez, A. (2004). R2O,
an extensible and semantically based database-to-
ontology mapping language. In: Proceedings of the
Second Workshop on Semantic Web and Databases,
Springer-Verlag, Berlin, Germany, pp. 1069-1070.
ISBN 978-3-540-24576-6.
Bizer, C. (2003). D2R MAP - A Database to RDF
Mapping Language. The Twelfth International World
Wide Web Conference, Budapest, Hungary.
Ghawi, R. and Cullot, N. (2007). Database-to-Ontology
Mapping Generation for Semantic Interoperability, 3rd
International Workshop on Database Interoperability
(InterDB 2007), held in conjunction with VLDB 2007,
Vienna, Austria.
Konstantinou, N., Spantos, D-E., Chalas, M., Solidakis E.
and Mitrou, N. (2006). VisAVis: An Approach to an
Intermediate Layer between Ontologies and Relational
Database Contents. International Workshop on Web
Information Systems Modeling (WISM 2006),
Luxembourg.
Svihla, M. and Jelinek, I. (2005). The Database to RDF
Mapping Module for an Easy Semantic Extending of
Dynamic Web Sites. IADIS International Conference
WWW/Internet, Lisabon, Portugal.
IWONTO-SQLBASEDMAPPINGFROMRELATIONALDATABASESTOONTOLOGIES
433