[UseCase] HP-3 Data integration and query transformation

** HP-3   Data integration and query transformation

* Outline:

Spoo Corp maintains a number of separate databases describing
different aspects of it's operating resources. For example, for IT
resources there might be a configuration management database, a
financial accounting database, a contracts database detailing related
outsourced maintenance contracts and so on. These databases are
maintained by different groups for different reasons and so have
distinct, though relatable, information models.

An enterprise planner wishes to explore the consequences of a change
in outsourcing policy and its impact on costs and what line of
business operations would be affected. This requires ad hoc joins of
information across each of these databases.

We assume the data models are published and the data is accessible
over Spoo's intranet using the SPARQL access protocol.

Our planner builds a set of data mapping rules which map each of the
data sets into a common model sufficient for her purpose. In doing
this she builds upon data mapping rules published by other analysts
and planners. In practice the data models change and each problem has
slightly different requirements so there is always some fine tuning to
the mapping required.

She is then able to query the combined databases to explore the
consequences of different planning options.

* Implications

Whilst the databases are finite there are no guarantees of completeness
and the common data model will certainly be semi-structured. So, in
contrast to the message transformation cases there is no useful
natural unit to scope negation or defaults.

The rules express the data mapping but operationally they might not be
applied directly. In this use case it is likely that they would be
used to guide a rewrite of the top level query into a set of local
(SPARQL) queries.

Received on Friday, 2 December 2005 11:33:13 UTC