Adoption of URIs as a Standard for Identifiers within ITER

This proposal starts from the realisation that CODAC signal names are just a special case of ITER identifiers in general. This proposal suggests that ITER decides on an organisational standard for identifiers (and their management) as soon as possible.

The proposal is to adopt a well defined subset of the recognised standard for Universal Resource Identifiers (shortened as URIs). For URI reference documentation see RFC 3986.

It is proposed that the URI subset, which is adopted by ITER, has URIs which start with http://www.iter.org/ns/, where ns stands for namespace.

The main advantage with the adoption of the URI standard is that the defined identifiers are globally unambiguous. At the same time their hierarchical structure suggest an intutive way to partition authority over identifier allocation in different areas of responsibility within ITER.

The allocation of names (URI-parts) in this top level ns namespace should be managed under the authority of the Head of IT at ITER. Each allocated URI-part in this top level namespace would denote a subordinate namespace, i.e. a major naming area such as CODAC signals. Each such subordinate namespace should be managed by a specified role within the ITER organisation. It is up to each such namespace manager to create further subordinate namespaces with delegated authority to specific roles for their management and so on.

Following this proposal it is further proposed that all ITER CODAC signals have URIs which begin with http://www.iter.org/ns/signal/ (the URI-part signal being subject to the above approval process).

Examples of proposed CODAC signal names:

http://www.iter.org/ns/signal/magnetics/plasma_current
http://www.iter.org/ns/signal/neutrons/14mev_neutron_flux

where magnetics and neutrons are assumed to be plant subsystems.

Adoption of Qualified Names as a Shorter URI Form

A major drawback with fully specified URIs is that they are too long. In order to address this problem it is proposed that ITER adopts the standard of Qualified Names or Qnames. Qnames were introduced by W3C into the XML document standard to solve this very problem.

Qnames are formal abbreviations of URIs. They are formed by prefixing a relative URI (i.e. a URI where many leading parts of the URI has been omitted - to be deduced by the context) with a defined name qualifier.

For example if a name qualifier is defined as

signal: = http://www.iter.org/ns/signal/

then the above CODAC signal examples can be written as

signal:magnetics/plasma_current
signal:neutrons/14mev_neutron_flux

In the XML standard the scope of a name qualifier is limited to the XML document. However, for an organisation like ITER it would make much sense to fully standardise name qualifiers across the organisation. This would allow for relatively short unambiguous identifiers to make references across all ITER documentation.

Moreover, standard qnames could be detected by web applications, which could automatically generate suitable hyperlinks to access the relevant metadata for the resource identified by each URI.

Note that according to the URI standard there is no promise that a URI specifies web access to the identified resource (even if the http scheme suggests so). In fact it is mostly inappropriate to provide such direct access. However, web access to the corresponding metadata is always recommended. The metadata web page may in turn provide a hyperlink to such direct resource access.

Allocation of Identifiers in Managed Namespaces

Managed namespaces as those discussed above can have their URIs allocated in two distinctly different manners:

• Managed name allocation - i.e. the organisational role, that is responsible for the namespace, performs allocation of suitable names.
• Name allocation through automatic mapping from another managed namespace.

The latter process can be distinctly subdivided into

• uri2uri mapping, which is easiest described as the replacement of one URI name qualifier with another.
• key2uri mapping, where any unique name key is prefixed by a URI name qualifier.

The key2uri mapping process means that it is always possible to retrofit any well managed namespace as a mapped URI (as is demonstrated by the CODAS CNS product).

Mapping of namespaces is a natural consequence of design and development. For example the refinement of a generic design item (identified by a URI) should probably be mapped into a differnet URI since a particular refinement is distinctly different from the original more generic design, but the new URI could share some elements from the URI of the generic design.

Sverker Griph, CODAS, JET