Information Technology Reference
In-Depth Information
Of course any piece of Representation Information could be domain specific,
but that content is not relevant to the Registry/Repository. It is important to note
that there may be multiple ways to describe something. For example Structure-type
Representation Information may come in the form of an EAST description, or a
DRB description or a DFDL description. All these are valid and each of these in
turn will have its own Representation Information.
In addition, it is possible that two archives may have identical copies of a piece of
data but may provide entirely separate pieces of Representation Information. This
is in many ways a duplication of effort. However the Registry/Repository will be
entirely unaware of this duplication since (1) it does not have a link back to the data,
as this would not be maintainable and (2) the pieces of Representation Information
are opaque binary objects as far as it is concerned.
A separate, value added, service may be developed by analysing the links
between data and Representation Information, in a way analogous to the ranking
algorithm used by Google. Such a service would enable one to say, for example,
that 99% of all archives use CPIDYYY as the Representation Information for a cer-
tain type of data. Such a statistic may influence others to use that particular piece
of Representation Information rather than some other, competing, Representation
Information.
New versions of Representation may be created from time to time, to improve
usability or accuracy. The versioning must be controlled and it will prove useful
to distinguish between a unique identifier for a particular version and a logical
identifier for all versions of the Representation Information. Using the logical iden-
tifier should return the latest (and presumably the best) version, which will change
as new versions are created, whereas using the unique identifier, or, equivalently,
providing a specific version number, should always provide that specific piece of
Representation Information.
Representation Information may be cached, that is to say copies may, for con-
venience, be kept, in a variety of locations, including packaged with the Data
Object. Caching is a well known optimisation technique and the appropriate steps
must be taken to ensure that the cache copies are identical with the original,
however the task is made easier because a particular piece of Representation
Information is never changed,
instead, as discussed above, a new version is
created.
16.2.1.2 Orchestration
The Orchestration component has to:
•
allow individuals to register their interests and expertise
•
collect information from (anonymous or registered) individuals about changes
in software, hardware, environment or Knowledge Base of any Designated
Community. This information will be passed on to the RepInfo Gap Manager
component.
•
receive information from the RepInfo Gap Manager component about a gap
which has been identified
