Databases Reference
In-Depth Information
We finally note that some of the partition-based and reuse-based match appro-
aches discussed above dealt with multiple subschemas, so they also implement some
form of
-way schema matching. An important building block in such advanced
match strategies is to search a collection of
n
(sub) schemas for the schema that is
most similar to a given schema. There are many other applications for the schema
search problem, e.g., finding similar peer schemas in P2P data integration or the
discovery of suitable web services (
Dong et al. 2004
;
Algergawy et al. 2010
).
n
4
Selected Match Systems
To further illustrate the state of the art, we discuss in this section the schema
matching capabilities in commercial tools as well as in selected research proto-
types. For better comparability, we restrict ourselves on systems for pairwise schema
matching.
4.1
Commercial Match Tools
In commercial tools, schema matching is typically a first step for generating exe-
cutable mappings (e.g., for data transformation) between schemas, particularly
XML schemas or relational database schemas. Systems such as IBM Infosphere
Data Architect, Microsoft Biztalk server, SAP Netweaver Process Integration, or
Altova MapForce provide a GUI-based mapping editor but still require a largely
manual specification of the match correspondences. In recent years, support for
automatic matching has improved and all mentioned systems can present users
equally named schema elements (typically within preselected schema fragments) as
match candidates. The Infosphere mapping editor also supports approximate name
matching and the use of external thesauri for linguistic matching. The mapping tool
of Microsoft Biztalk server 2010 has significantly improved for better matching
large schemas (
www.microsoft.com/biztalk
). It supports an enhanced user interface
to better visualize complex mappings similar as described in
Bernstein et al.
(
2006
).
Furthermore, it supports approximate name matching by a new search functionality
called “indicative matching.”
The increasing support in commercial tools underlines the high practical impor-
tance of automatic schema matching. However, the tools need much further improv-
ement to reduce the manual mapping effort especially for large match tasks. For
example, commercial tools do neither support structural matching nor any of the
advanced techniques discussed in Sect.
3
.
