Databases Reference
In-Depth Information
combined with the results of other matchers) in order to obtain a complete schema
mapping.
Saha et al.
(
2010
) focuses on the second step and describes an approach
called
schema covering
to partition the input schemas such that the partitions can
be matched to schema fragments in the repository. They first apply a filter step
to determine relevant repository fragments with some similarity with the schemas
to be matched. Then for each of the remaining repository fragments, the maximal
subgraphs in the schemas are determined that can be covered by the fragment.
To speed-up the similarity computations, filtering and subgraph identification uti-
lize a predetermined index of the schema element names and their positions in the
repository schemas and the schemas to be matched.
SAP also works on an ambitious project called Warp10 to exploit the reuse of
XML schemas and mappings for business integration including improved schema
matching (
SAP 2010
). As indicated in Fig.
1.5
, the key idea is to maintain a
central repository maintaining a global schema (called consolidated data model)
that includes semantically consolidated versions of individual schemas or schema
fragments. Consolidation is based on the UN/CEFACT core component technical
specification (CCTS) rules for uniformly naming and structuring concepts. The
global schema is initially populated by standard business schemas and data types
(e.g., from SAP) and can be semiautomatically and collaboratively extended by
integrating specific schemas. The correct integration of such schemas is likely a
complex merge operation needing the control by domain experts. Once mappings
between schemas and the global schema are established, they can be reused for
quickly solving new integration tasks, particularly for matching between schemas
(to match schemas
S1
with
S2
, one has to compose their mappings with the global
schema
G, S1-G
and
G-S2
). Unfortunately, the details about how the global schema
is maintained are not yet published.
Fig. 1.5
Mappings between schemas and the consolidated data model in Warp10 (from
SAP
(
2010
))
