Explore GeoScience Ontology for use in GeoKB

[skybristol]

I need to dig into the GeoScience Ontology to work out how it blends with what we already have and use it to fill in blanks as a primary source. The biggest thing in this is working through the mapping of the formal ontology structure to the slightly less formal Wikibase model.

Source is on GitHub here - https://github.com/Loop3D/

[skybristol]

After working through the upper level ontologies and seeing how this is all laid out, I think we will be able to incorporate the vast majority of the GSO into the GeoKB, sticking close to its formal ontology representation but giving us flexibility in terms of adding additional claims as needed. I've prepped a spreadsheet for review of the rock types from the TTL source:

https://github.com/Loop3D/GKM/blob/master/Loop3D-GSO/Modules/GSO-Geologic_Rock_Material.ttl

I'll pass this on to our geoscientists for review and comment. I'm particularly interested in the inclusion of individual reference sources to see how we might incorporate these in an explicit way.

[skybristol]

Some initial thoughts on GeoKB mapping focused on the initial case of reworking our rock type classification:

• Multiple labels on items should fit easily enough into the simple alias/alt-label concept in Wikibase
• For things like the rock material classification that includes original source references, it would be nice to trace these out to items representing those sources. I'll need to decide if we want to simply build this iteratively, starting with stubs.
• For the rock classification, I will likely "replace" what we have now from Mindat, which some on our team have indicated is overly complex and lacking in original source links to references the group is already comfortable using. We will figure out the connections to Mindat (many of which are encoded into the GSO work) so we can retain the sameAs relationships.
• Comments in the rock classification have quite rich content. We may do some kind of standardized description or leave in the short descriptions from Mindat and drop the full comments into item talk pages.

[skybristol]

One of the bigger things to work out here was the Common Core Ontologies basis of the GSO and how that relates to the Wikibase framework, which is not based on strict semantic modeling or the Web Ontology Language (OWL). This whitepaper from a couple years ago helped in some respects. As stated in other discussions, I'm trying to balance pragmatism with explicit modeling of concepts in a way that encourages broad-based participation in building the knowledgebase as opposed to needing overly cumbersome ontology engineering just to get knowledge captured. We want a little more rigor than is evidenced in Wikidata but we want a framework that encourages everyone we let in the front door to contribute. Part of that comes from laying down an appropriate foundation with well-constructed ontologies like the GSO so that there are many of the appropriately modeled items to link to.