Databases Reference
In-Depth Information
In many areas of science, researchers are investigating how best to pair human
input with automated procedures. For example, in the area of intelligent robot
design, some researchers believe that the future of the field lies not in the devel-
opment of fully automated robots, but in the development of partially automated
ones [
Coradeschi and Saffiotti 2006
]. Some tasks, such as classification and pattern
recognition, are very difficult and robots need help from humans in performing these
tasks. At the same time, robots can help humans with tedious and repetitive tasks.
Similarly, in ontology matching, humans have access to vast amounts of background
knowledge, which they can use to help make inductive judgments about potential
correspondences.
In general, potential matching correspondences produced by a matching tool
must be examined by a domain or ontology expert. The expert must determine
the correspondences that are correct, remove false positives, and create additional
correspondences missed by the automated procedure. This process is both time con-
suming and cognitively demanding. It requires understanding of both ontologies that
are being mapped and how they relate to each other. Furthermore, both the ontolo-
gies and the number of candidate matching correspondences that the tools produce
can be very large. Researchers have largely focused on improving the performance
of the algorithms themselves. However, recently there has been a growing trend
toward a more human-centered approach to ontology matching.
Examining and supporting the symbiosis between tool and user has been gaining
more prominence and more tools that support a
semiautomatic
process are becom-
ing available. Shvaiko et al. discuss ten challenges for ontology matching, three of
which directly relate to the user:
user involvement
,
explanation of matching results
,
and
social and collaborative ontology matching
[
Shvaiko and Euzenat 2008
]. One
approach researchers have been exploring to help support user involvement is infor-
mation visualization techniques, such as those used by AlViz [
Lanzenberger and
Sampson
2006
]andC
OG
Z[
Falconer and Storey 2007b
]. The
International Work-
shop on Ontology Alignment and Visualization
1
was created as a platform for
researchers to share and explore new visual techniques to support the matching
process. Another growing trend is the use of Web 2.0 approaches to help support
the social and collaborative matching process. Researchers are exploring the util-
ity of
crowdsourcing
to help facilitate the process of generating many matching
correspondences [
Noy et al. 2008
;
Zhdanova 2005
].
These new trends in ontology matching research offer an exciting and interesting
alternative to completely manual or completely automated processes. The research
emphasis is shifting. New research is investigating how to gain a better understand-
ing of the cognitive demands placed on the user during a matching procedure, how
communities of users can work together to create more comprehensive and precise
matchings, and how to make the most effective use of automation. Research on
these topic areas is still in its infancy, but the future of the field lies in a joint effort
between human and machine.
1
http://www.ifs.tuwien.ac.at/?mlanzenberger/OnAV10/
.
