ISO 15926 TUTORIAL - LESSON 4 - Reference Ontologies

In this lesson I will address the

Reference Ontologies

Based on both Upper Ontologies we now define instances of a metaclass (here: ClassOfInanimatePhysicalObject) that are also a subclassses of a class(here: InanimatePhysicalObject), marked in red.

The Reference Ontology probably is to be in some next version of the RDL (Reference Data Library).

Here is a brief history of the RDL (which you may skip and forget):

• Once upon a time we were working on ISO 10303-221 - 'Functional data and their schematic representation for process plants'. In that 'AP221', as it was dubbed, was an Annex M with reference data;
• We then set up StepLib;
• The folks in Norway initiated their RDS (Reference Data System), now called the PoscCaesar RDL, and harmonized that with StepLib;
• ISO 15926-4 saw the light in 2007, in spreadsheet format, so not processable (there are some hurdles to take in ISO);
• RDL grew tremendously from contributions of some 70 companies, but as is often the case with rapid growth it got a bit in disarray;
• Onno Paap and myself reorganized the RDL, guided by a group called MRAIL (set up by Robin Benjamins of Bechtel) and created the endpoint http://data.15926.org/rdl/ dubbed 'RDL2';
• PoscCaesar is updating their RDL based on the RDL2 contents, which will be launched shortly, adding a lot of features;
• RDL2 will, for the foreseeable future, be used for additions and revisions, the results of which are uploaded to the PoscCaesar RDL;
• Meanwhile the CFIHOS Project, first under the auspicies of USPI and later of IOGP (as their JIP36 Project) created their own RDL, which has been mapped by us as an extension of RDL2;
• We harmonized the RDL2 with the DEXPI reference data and with the IEC CDD instrumentation data;
• We started updating ISO 15926-4 with the results of 14 years since its inception. This will be a longdrawn process because this requires ISO ballots (which is fine, no misunderstanding about that).

BUT, we aren't done then, because what we now have is:

• a taxonomy with a tree structure of classes, and not yet
  
• an ontology that defines relationships between the classes in that taxonomy.

Building such a Reference Ontology for thousands of classes is a Herculean task, but as the proverb says: "Even the longest journey takes a first step". To do this in a topdown mode requires a CFIHOS-like project with generous funding. As a matter of fact they intend to build a meronomy of part-whole hierarchies, and add their (minimum) requirements for functional and technical data, where the IOGP JIP33 Project defines documentation requirements. But keep in mind that, large as the IOGP world is, it is not representative for the entire global process industries.

So a lot of contributions will have to come in a bottom up mode.

As already mentioned, the RDL has extensions with local reference data, that jointly form a hierarchy that is spread over many stores, but tightly integrated. To illustrate that this picture:

Every company or project can set up their own RDL extension. It would be great when all parties in the process industries would do that, for global interoperability reasons. See also this topic.

The typical example in http://data.15926.org/cfihos/30000295 shows how that is done for the CFIHOS RDL extension.
Of course their will be differences of opinion (engineers seldomly fully agree with each other), leading to different taxonomies and hence different inheritances.
Those inheritances becomes very important when we turn the Reference Library into a Reference Ontology.

What's in it?

At present this can be found in the RDL2 :