Open nicholascar opened 4 years ago
@dr-shorthair @lewismc @smrgeoinfo any chance of a review here? There are a lot of classes, sure, but they are all done in the same style so if you review a few and like them, you should be happy with the rest.
Hi Simon,
We will soon find out!!
Cheers,
John [removing email details - Nick]
@dr-shorthair what's your response to @johnmckellar's comments
@smrgeoinfo @lewismc can you please review? I'm keen to keep momentum here! Let's get PRs merged as they can always be revised.
I have another PR of a comprehensive sphere (atmosphere, hydrosphere) updates ready...
My concern is primarily around the position of Geologic Province in the hierarchy, and also whether they can be nested. I'm looking for some feedback from non-Australians as to whether this conception is local or general.
Is there a diagram of the subsumption hierarchy? That would also help in the evaluation.
First 1/3 of hierarchy
Second 1/3 of hierarchy:
Third 1/3 of heirarchy:
@dr-shorthair wrote:
I'm looking for some feedback from non-Australians as to whether this conception is local or general.
Note that John says "Much of the proposed structure of the Features ontology is based on the Guide for Geological Nomenclature in Sweden" (as per https://github.com/ESIPFed/sweet/pull/187#discussion_r391337771)
Yes, but presenting a single tree is misleading as it does not tease out the relationships between individuals. Most of this is fine, but I'm concerned that in some places there is a conflation of
These are very different relationships.
In putting this in the vernacular, so a very simple beginner such as I can understand, I take it that you are saying that, for a term (in this case ‘trough’) that can have different applications and thus meanings (and our languages are full of such terms), we need to explain, and inter-relate them all (and indeed compare them where no relationships exist)?? Does not the context for usage of the term provide this, here in the hierarchy/tree? Trough type can be further subdivided geologically or tectonically in the hierarchy? What do you think Nick, Derek?
tease out the relationships between individuals
@johnmckellar and I have already discussed adding some more properties to capture the various rankings of subclasses within a superclass (LithostratigraphicUnit subclasses for instance) and I agree, we will likely need to look into breaking some other subclass relations into part/whole.
If you can suggest sensible part/whole relations, I'll add them into the PR. Do you have some obvious ones in mind?
dcterms:isPartOf
dcterms:hasPart
To start with anyway
.. position of Geologic Province in the hierarchy,...> Is there a diagram of the subsumption hierarchy? That would also help in the evaluation.
It seems to me that GeolgoicProvinces are defined by shared history and ?perhaps by spatial contiguity during that history. Thus not necessarily a subclass of soreag:TectonicEntity. I don't see what the distinction is between soreag:TectonicEntity and soreag:GeologicProvince is anyway.
dcterms:isPartOf
dcterms:hasPart
To start with anyway
In SWEET context probably better to use sorel:partOf
and sorel:contains
AFAICT there are two things being addressed in this proposal and reviews:
textual definitions, which optimally should reflect the community usage and consensus, else if this is not yet agreed then definitions must be accompanied by their provenance
relationships between classes.
I was primarily concerned that the latter was expressed purely as a subsumption hierarchy, even though some of the relationships should be related to partonymy.
SWEET does have a set of high-level relations available though I don't think they are used much. To support the geological features hierarchy I would expect to see these appear in relationships like
soreag:SubBasin rdfs:subClassOf [
a owl:Restriction ;
owl:allValuesFrom soreag:Basin ;
owl:onProperty sorel:partOf ;
] ;
.
I'm now finding it difficult to keep on top of all the separate discussions here. I suggest breaking it out into multiple issues. The PR was very big and now looks like it should be broken up into a few pieces, matched with issue threads.
I'm now finding it difficult to keep on top of all the separate discussions here. I suggest breaking it out into multiple issues. The PR was very big and now looks like it should be broken up into a few pieces, matched with issue threads.
I'm keen not to break the PR up. With the decision to focus on textual definitions first and to tidy the subsumption and part/whole relations second and to stick to those relations only, the main work here, going on at points above, is to address the definitional challenges and that's all a single conceptual thing. I think we are close to satisfying the points above too.
I will make a series of changes to the PR shortly with the updating commits referencing comments here. If those are accepted, I'll then close comments above which will neaten the PR and perhaps make it possible to pass.
I really think we should set up a Semantic Harmonization Cluster for geological features. It helped us work through many of the same convoluted issues for the cryosphere (@rduerr). Happy to bring in the ENVO semantic analysis approach, which helped unpack things.
The deep-thinkers in geological ontologisation (Steve Richard, Boyan Brodaric, Mark Jessell, others) are mostly working in the Loop3D project, the results of which are unfortunately not open, yet (I think about 12 months overdue now). This is frustrating, but I think it would be risky to start another process in parallel.
This is a major update to SWEET Ontology Realm Geologic ontology and a couple of related ontologies.
Within SWEET Ontology Realm Geologic, it provides a class hierarchy beneath the class
soreag:GeologicFeature
of about 50 classes, all with definitions drawn from literature which is cited. It also provides cited definitions for a few classes in the SWEET Ontology Realm Geologic Basin, SWEET Ontology Realm Geologic Continental & SWEET Ontology Realm Geologic Orogen ontologies.It tidies up the top-level hierarchy of SWEET Ontology Realm Geologic too by adding in explicit subclassing triples for classes for which these are inferred by reasoning. This helps with hierarchical display in tools such as Protege.
All changes are attributed to the Geological Survey of Queensland, Nicholas Car and John McKellar equally since John & I did the work for the GSQ.
I was able to provide pleasant diffs for:
since changes were small but I wasn't able to do so for the main realmGeol.ttl file since Protege reformatting moved things around (e.g. new spaces).