202 9 The Transformation Process
Using such a tool requires significant manual effort to correct wrong matches
and add missing matches. In practice, schema matching is done manually
by domain experts, and is very time-consuming when there are many data
sources or when schemas are large or complex.
Automated ontology mediation systems are designed to reduce manual
effort. However, such a system requires a substantial amount of time to
prepare input to the system as well as to guide the matching process. This
amount of time can be substantial, and may easily swamp the amount of time
saved by using the system. Unfortunately, existing schema-matching sys-
tems focus on measuring accuracy and completeness rather than on whether
they provide a net gain. Schema-matching systems have now been proposed
(Wang et al. 2004) that address this issue. However, such systems are not yet
available. The best that one can hope for from current systems is that they
can help one to record and to manage the schema matches that have been
detected, by whatever means.
One example of a schema integration tool is COMA, developed at the Uni-
versity of Leipzig (Do and Rahm 2002; Do et al. 2002), but there are many
others. See (Rahm and Bernstein 2001) for a survey of these tools. Some of
these tools also deal with XML DTDs (Nam et al. 2002). Unfortunately, they
are only research prototypes and do not seem to be available for download-
ing.
There are many ontology mediation projects, and some have developed
prototypes, such as PROMPT (Noy and Musen 2000) from the Stanford Med-
ical Informatics laboratory and the Semantic Knowledge Articulation Tool
(SKAT), also from Stanford (Mitra et al. 1999), but as with schema integra-
tion, none seem to be available for public use, either via open source software
or commercial software.Summary
- Reconciling differing terminology has many names depending on the par-
ticular context where it is done, such as: ontology mediation, schema inte-
gration, data warehousing, virtual data integration, query discovery, and
schema matching. - Automated ontology mediation systems attempt to reduce manual effort,
but they rarely provide a net gain. - Most automated ontology mediation systems are still research prototypes.