OWL Ontologies - equivalent class expressions

[Michel Dumontier]

Hi,
There is no value to, and it is expensive to reason over, NamedClass
(es) Equivalent to the Parent + "hasType" some NamedClass. These
axioms need to be removed. Also remove hasType from the ontology -
hierarchical typing in OWL is down via rdfs:subClassOf relations. In
OWL, you can just make unions or intersections of classes if that is
what you need to express, although that doesn't appear to be the case
in the ontologies.

[chem...@googlemail.com]

Hi Michel,

what's your alternative? Do you want me to assert multiple parentage
instead?

N.

On Apr 6, 9:34 pm, Michel Dumontier <michel.dumont...@gmail.com>
wrote:

[Michel Dumontier]

A good OWL ontology should provide an explicit description of the necessary (and possibly sufficient) attributes and relationships between the named entities and/or datatype values. If the features are explicitly modeled, then there is no need to build the disjoint hierarchy. A disjoint hierarchy is necessary if you make arbitrary decisions about what makes one entity different than another. These kinds of ontologies are rampant in biology. Instead, you accurately describe the entity by placing class restrictions on it. The reasoner will infer the hierarchy on your behalf and also identify inconsistencies automatically. 

Let's take metrology ontology first.

I see that "NamedTechnique" is a parent for (all?) experimental technique classes, but this isn't necessary. By default a class in the ontology is named, and if they are a type of experimental technique, then they are named.

GPC_DRI is asserted to be "has type" some "Chemical Analysis Technique", and "Gel permeation chromatography". What is the relationship between these types? While all chemical analyses techniques are not gel permeation chromatography, are all gel permeation chromatography types of chemical analysis techniques? If so, what makes the latter true? Is it that it a chemical analysis technique involves a chemical, and so does a gel permeation chromatography? This kind of explicit modeling of what is involved will help you autogenerate the hierarchy.

Since this ontology is looking at ways to investigate chemicals, i would suggest you have a look at the ontology of biomedical investigation: http://obi-ontology.org/page/Main_Page

they have worked out a number of issues with respect to instrumentation.

-=Michel=-

--
Michel Dumontier
Assistant Professor of Bioinformatics
http://dumontierlab.com

[Michel Dumontier]

As an additional note on the metrology ontology, the class restrictions involving 'has instrument' are incorrect. Don't use hasValue - use an existential restriction (e.g. some). that way you know that an instance of the technique will involve some real world instrument, instead of there only being one such instrument in the world.

-=Michel=-