Produce IRIs for classification scheme datatypes

[cspayne]

We need to generate a set of IRIs for the datatype of each classification scheme that is in the LC list of classification schemes.

Gordon has offered to help create these datasets, but first we need to decide where we storing these IRIs, sooner rather than later. Current options discussed have been publishing them as a part of UW Libraries semantic web data, the same way we have with 006, 007, and 008 vocabularies, or in Wikidata.

Comments pulled from 050 issue

[cspayne]

https://github.com/uwlib-cams/MARC2RDA/issues/82#issuecomment-1997331641 @GordonDunsire:

... We need to generate a set of IRIs for the datatype of each classification scheme that is in the [LC list of classification schemes] (https://id.loc.gov/vocabulary/classSchemes.html). This should be added to the other datasets and value vocabularies that UW is managing for the transform. The datatype URIs can be of the form UWbase/classSchemesLC/code-for-scheme/

@CECSpecialistI: I think it is best to generate this dataset (of datatypes) as a pre-set for the transform, rather than when the transform is run. The dataset is fixed (until LC adds a new scheme), and only codes from the LC list are valid data for the transform. This overlaps with the discussion on IRI construction for RDA entity datasets, SKOS value vocabularies for classification numbers and subject headings, and the pattern already established for the MARC 21 value vocabularies. Is Wikidata a better place for these pre-transform vocabularies?

[cspayne]

https://github.com/uwlib-cams/MARC2RDA/issues/82#issuecomment-1998473086 @CECSpecialistI:

@GordonDunsire I think you're probably right. I'm not sure whether Wikidata wants these vocabularies, or if we ought to publish them as part of the UW Libraries semantic web data as we have done with the OMR vocabularies. We should talk about our options and, of course, about the labor it will take to get it done (something we find ourselves shorter on than we were a few months ago I'm afraid)

[cspayne]

https://github.com/uwlib-cams/MARC2RDA/issues/82#issuecomment-1999317412 @GordonDunsire :

@CECSpecialistI: I can help with creating these datasets. The process is roughly:

1. Scrape the LC list (in text or RDF/XML to extract the name of the scheme and its code.
2. Use a spreadsheet formula to output our local RDF/Turtle statements.
3. Use an online RDF converter to reformat the output into the format required for import into a namespace/dataset.

I can do some prep now, and when we decide on the IRI templates I can apply them to the spreadsheets.

[cspayne]

https://github.com/uwlib-cams/MARC2RDA/issues/82#issuecomment-1999851333 @CECSpecialistI :

Thank you Gordon! We have been using a DOI/Datacite process that is fairly laborious, but if we decide on Wikidata that would be less labor-intensive. I appreciate your willingness to help with this very much!

https://github.com/uwlib-cams/MARC2RDA/issues/82#issuecomment-2023523987 @CECSpecialistI :

@cspayne @pennylenger @GordonDunsire deciding on where these values belong is probably something we need to consider in the next few weeks. Any thoughts on Wikidata vs. DOI/Datacite/UW Semantic Web Data? Wikidata may be more reusable.

https://github.com/uwlib-cams/MARC2RDA/issues/82#issuecomment-2023771221 @cspayne :

If we use Wikidata, will we still be producing a dataset that lists these values? For the transform I'll need a document or dataset to traverse through and pull IRIs from.

https://github.com/uwlib-cams/MARC2RDA/issues/82#issuecomment-2023892290 @CECSpecialistI :

Yes, it would be part of the Wikidata graph in that case.

[cspayne]

https://github.com/uwlib-cams/MARC2RDA/issues/82#issuecomment-2039445459 @GordonDunsire :

@cspayne and others: as @pennylenger says, SchemeNotationLCC is a datatype for the syntax of an LCC classification number. The datatype is described by documentation such as Library of Congress Classification; for example, the main categories or types of LCC number begin with a 1, 2, or 3 upper-case letter, and specific topics with a category are encoded with a 1-4 digit 'number'. We will mint the datatypes in order to follow good practice outlined in the SKOS primer, but we will not link the datatype to any documentation (that's a whole separate project :-).

See the SKOS primer section 4.6 Notations: 'The first, preferred technique is to use the skos:notation property. This property allows a concept to be attached to an RDF typed literal—a literal with an explicit datatype [RDF-PRIMER]. The datatype of the literal specifies a syntax encoding scheme, which fits the usage of notations in the concerned KOS.' There is more technical information in SKOS reference section 6.5.1.

[GordonDunsire]

@cspayne, @CECSpecialistI: One factor in favour of Wikidata is that these datatypes are globally applied; that is, whenever a standard classification scheme is expressed in SKOS and used in linked open data. [The MARC code vocabularies, on the other hand, are local to NDMSO/LC and MARC 21.] However, we should check if Wikidata can, and wants to, store datatypes for strings.

[CECSpecialistI]

I'm having trouble understanding what these IRIs would stand for and how they would differ from the ones LC has already published at id.loc. If we have a clear idea of what they are an "instance of", the id.loc identifier for them, then I can create Wikidata items for them using Quickstatements relatively easily.

[GordonDunsire]

@CECSpecialistI: These are string datatypes and not the schemes that use the datatypes, so they have different IRIs.

I think they are an instance of https://www.wikidata.org/wiki/Q31385480

This is (ultimately) a subclass of LRM Manifestation, which I think is correct. I'm not sure how to search Wikidata to see if there are any instances of datatype already established.

[CECSpecialistI]

Here's a query that finds the instances of datatype in Wikidata. Wikidata has two of these, one for rdf:HTML, and one for rdf:XMLLiteral. I think we could add more string datatypes to Wikidata. If someone wants to generate the list of datasets and what their classification schemes are, I can help with the Wikidata aspect at that point.

[GordonDunsire]

@CECSpecialistI: How is the data to be loaded to Wikidata? Manually? Batch upload?

We already have an RDF dataset for the LC classification schemes at id.loc.gov. We can process this file to create an RDF dataset for the corresponding datatypes. For example, if WD allows import and retention of local IRIs, we change the base domain of each classification scheme from 'http://id.loc.gov/vocabulary/classSchemes/' to, say, 'http:/localUW/classSchemes/datatypes/'; this is real easy to do. We can extract the scheme label/title and replace 'http://www.loc.gov/mads/rdf/v1#authoritativeLabel' with an appropriate WD label element for each dataset., etc.

We only need to process one statement for each scheme from the LC dataset, for example:

'<http://id.loc.gov/vocabulary/classSchemes/msdocs> <http://www.loc.gov/mads/rdf/v1#authoritativeLabel> "Mississippi state government publications. Vol. 1, July 1975/June 1980" .

If a manual transfer is used, we can just copy/paste the labels and associated class scheme IRIs from the LC html page.

I think it is ok to collapse the DDC editions in the generic DDC entry; that is, ignore the DDC edition schemes. This is because the syntax of a DDC notation is the same for all editions. I suspect that there are other distinct schemes that use the same syntax, for example schemes derived from LC Classification, but it will take too long to check each one.

[CECSpecialistI]

I would use a batch upload for Wikidata process for Wikidata based on a spreadsheet. I'd probably use Quickstatements V1 for this. https://quickstatements.toolforge.org/#/

[CECSpecialistI]

Or maybe OpenRefine's Wikidata plugin

[AdamSchiff]

Is this also a datatype: Q31381203 (http://www.wikidata.org/entity/Q31381203)

Adam L. Schiff Principal Cataloger University of Washington Libraries (206) 543-8409 @.***

---

From: Crystal Yragui @.> Sent: Tuesday, September 17, 2024 5:36 PM To: uwlib-cams/MARC2RDA @.> Cc: Subscribed @.***> Subject: Re: [uwlib-cams/MARC2RDA] Produce IRIs for classification scheme datatypes (Issue #465)

Here's a queryhttps://urldefense.com/v3/__https://w.wiki/BE3i__;!!K-Hz7m0Vt54!g5gb6sXADH4PeyqcZcytKlDDLOvMce6n_psrh7XqQ9wuDwJj_6-MDkvWvdXR7osY-tige2N8yrKDVtZOTuk6cqs$ that finds the instances of datatype in Wikidata. Wikidata has two of these, one for rdf:HTMLhttps://urldefense.com/v3/__https://www.wikidata.org/wiki/Q31474804__;!!K-Hz7m0Vt54!g5gb6sXADH4PeyqcZcytKlDDLOvMce6n_psrh7XqQ9wuDwJj_6-MDkvWvdXR7osY-tige2N8yrKDVtZOMnPprfA$, and one for rdf:XMLLiteralhttps://urldefense.com/v3/__https://www.wikidata.org/wiki/Q31474952__;!!K-Hz7m0Vt54!g5gb6sXADH4PeyqcZcytKlDDLOvMce6n_psrh7XqQ9wuDwJj_6-MDkvWvdXR7osY-tige2N8yrKDVtZOg5-sEaA$. I think we could add more string datatypes to Wikidata. If someone wants to generate the list of datasets and what their classification schemes are, I can help with the Wikidata aspect at that point.

— Reply to this email directly, view it on GitHubhttps://urldefense.com/v3/__https://github.com/uwlib-cams/MARC2RDA/issues/465*issuecomment-2357258722__;Iw!!K-Hz7m0Vt54!g5gb6sXADH4PeyqcZcytKlDDLOvMce6n_psrh7XqQ9wuDwJj_6-MDkvWvdXR7osY-tige2N8yrKDVtZOJqpqL5Q$, or unsubscribehttps://urldefense.com/v3/__https://github.com/notifications/unsubscribe-auth/ADFBVBZFZZK6CNH3B6YFB23ZXDDJ7AVCNFSM6AAAAABNM3TFKKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGNJXGI2TQNZSGI__;!!K-Hz7m0Vt54!g5gb6sXADH4PeyqcZcytKlDDLOvMce6n_psrh7XqQ9wuDwJj_6-MDkvWvdXR7osY-tige2N8yrKDVtZO6QWQeyY$. You are receiving this because you are subscribed to this thread.Message ID: @.***>

[GordonDunsire]

@AdamSchiff: Yes. An RDF literal is not just a string, but has additional syntactical structures, such as the xml language attribute and a datatype IRI. The transform does not need a datatype for its string values in general because there are no syntax constraints on the content.

But this prompts me to think that, in addition to datatypes for classification notations to comply with SKOS best practice, it would be good to add a 'MARC exchange format' datatype to the full MARC record that is to be included in the transform output.

@CECSpecialistI: can you add such a datatype to Wikidata, with a WD 'described by' or 'source' property that points to the ISO standard that is applied to MARC exchange format, unless there is a specific MARC exchange format description somewhere (online, with a URL/IRI).

[CECSpecialistI]

Sure, I can add anything to Wikidata if there's a description for it online that you can point me to. I'm compiling the Quickstatements spreadsheet for the LOC list right now. The output from the list is (of course) not cooperating well with OpenRefine or Quickstatements so it's needing a lot of manipulation so it's ok if it takes a little while to track down.

[CECSpecialistI]

IRIs for classification scheme datatypes have been produced: https://w.wiki/BFBw

[cspayne]

Yay! Thank you Crystal.

Two quick questions for @GordonDunsire:

1. For some 3XX fields such as 334, we mint concepts if the source is not rda. If there is a code provided along with a term in that field, the code is mapped to skos:notation for that concept. These sources come from the Genre/Form Code and Term list, and possibly from other lists as well. Does this mean we need to mint datatype IRIs from these lists as well? Or are we only doing this for concepts that come from classification schemes?

2. For schemes that are not in the list (such as UDC abridged), do we just not provide a datatype?

[GordonDunsire]

@CECSpecialistI: Great to see the project's datatypes linking up to the pre-existing Wikidata property LoC and MARC vocabularies ID!

@cspayne: 1. Yes, if we mint an instance of skos:notation it is good practice to type the literal value. I assume it won't cause a significant delay, and I expect providing the datatypes in WD will be big(ger) kudos to UW and the project.

2, We should provide a datatype for each 'known' scheme that we use. The WD entry cannot use the LoC and MARC property; can @CECSpecialistI find another?

[AdamSchiff]

Crystal,

Does it make sense to say that one of these is both an instance of RDF datatype IRIhttps://www.wikidata.org/wiki/Q31385480 AND a part of RDF datatype IRIhttps://www.wikidata.org/wiki/Q31385480? It seems a bit odd to me to use both of these properties with the same value.

Adam

Adam L. Schiff Principal Cataloger University of Washington Libraries (206) 543-8409 @.***

---

From: Crystal Yragui @.> Sent: Thursday, September 19, 2024 1:48 PM To: uwlib-cams/MARC2RDA @.> Cc: Adam L Schiff @.>; Mention @.> Subject: Re: [uwlib-cams/MARC2RDA] Produce IRIs for classification scheme datatypes (Issue #465)

IRIs for classification scheme datatypes have been produced: https://w.wiki/BFBwhttps://urldefense.com/v3/__https://w.wiki/BFBw__;!!K-Hz7m0Vt54!m7AC66lt4XreJuPvgk-yhnYVCaQjz7759lxryR0bD1vSUKme-ia_DIfJxy6BYfp0vURT3eU76irqw-A-on_ZE_A$

— Reply to this email directly, view it on GitHubhttps://urldefense.com/v3/__https://github.com/uwlib-cams/MARC2RDA/issues/465*issuecomment-2362144761__;Iw!!K-Hz7m0Vt54!m7AC66lt4XreJuPvgk-yhnYVCaQjz7759lxryR0bD1vSUKme-ia_DIfJxy6BYfp0vURT3eU76irqw-A-DTdmfbY$, or unsubscribehttps://urldefense.com/v3/__https://github.com/notifications/unsubscribe-auth/ADFBVB2KJVT62WUNJG63ZWDZXM2ATAVCNFSM6AAAAABNM3TFKKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGNRSGE2DINZWGE__;!!K-Hz7m0Vt54!m7AC66lt4XreJuPvgk-yhnYVCaQjz7759lxryR0bD1vSUKme-ia_DIfJxy6BYfp0vURT3eU76irqw-A-8gzBsPo$. You are receiving this because you were mentioned.Message ID: @.***>

[AdamSchiff]

I added another property to two of these to see if this seems appropriate for all of them.

See https://www.wikidata.org/wiki/Q130324885 and https://www.wikidata.org/wiki/Q130324891https://www.wikidata.org/wiki/Q130324891. In each I added the property "code":

[cid:810f38b9-4c0b-4e68-94da-21192a81fe09]

[cid:b04fc9c6-0fba-4682-9a57-243315dd4b0b]

Adam L. Schiff Principal Cataloger University of Washington Libraries (206) 543-8409 @.***

---

From: Crystal Yragui @.> Sent: Thursday, September 19, 2024 1:48 PM To: uwlib-cams/MARC2RDA @.> Cc: Adam L Schiff @.>; Mention @.> Subject: Re: [uwlib-cams/MARC2RDA] Produce IRIs for classification scheme datatypes (Issue #465)

IRIs for classification scheme datatypes have been produced: https://w.wiki/BFBwhttps://urldefense.com/v3/__https://w.wiki/BFBw__;!!K-Hz7m0Vt54!m7AC66lt4XreJuPvgk-yhnYVCaQjz7759lxryR0bD1vSUKme-ia_DIfJxy6BYfp0vURT3eU76irqw-A-on_ZE_A$

— Reply to this email directly, view it on GitHubhttps://urldefense.com/v3/__https://github.com/uwlib-cams/MARC2RDA/issues/465*issuecomment-2362144761__;Iw!!K-Hz7m0Vt54!m7AC66lt4XreJuPvgk-yhnYVCaQjz7759lxryR0bD1vSUKme-ia_DIfJxy6BYfp0vURT3eU76irqw-A-DTdmfbY$, or unsubscribehttps://urldefense.com/v3/__https://github.com/notifications/unsubscribe-auth/ADFBVB2KJVT62WUNJG63ZWDZXM2ATAVCNFSM6AAAAABNM3TFKKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGNRSGE2DINZWGE__;!!K-Hz7m0Vt54!m7AC66lt4XreJuPvgk-yhnYVCaQjz7759lxryR0bD1vSUKme-ia_DIfJxy6BYfp0vURT3eU76irqw-A-8gzBsPo$. You are receiving this because you were mentioned.Message ID: @.***>

[CECSpecialistI]

Crystal, Does it make sense to say that one of these is both an instance of RDF datatype IRIhttps://www.wikidata.org/wiki/Q31385480 AND a part of RDF datatype IRIhttps://www.wikidata.org/wiki/Q31385480? It seems a bit odd to me to use both of these properties with the same value. Adam Adam L. Schiff Principal Cataloger University of Washington Libraries (206) 543-8409 @. … ____ From: Crystal Yragui @.> Sent: Thursday, September 19, 2024 1:48 PM To: uwlib-cams/MARC2RDA @.> Cc: Adam L Schiff @.>; Mention @.> Subject: Re: [uwlib-cams/MARC2RDA] Produce IRIs for classification scheme datatypes (Issue #465) IRIs for classification scheme datatypes have been produced: https://w.wiki/BFBwhttps://urldefense.com/v3/__https://w.wiki/BFBw__;!!K-Hz7m0Vt54!m7AC66lt4XreJuPvgk-yhnYVCaQjz7759lxryR0bD1vSUKme-ia_DIfJxy6BYfp0vURT3eU76irqw-A-on_ZE_A$ — Reply to this email directly, view it on GitHub<https://urldefense.com/v3/__https://github.com/uwlib-cams/MARC2RDA/issues/465issuecomment-2362144761__;Iw!!K-Hz7m0Vt54!m7AC66lt4XreJuPvgk-yhnYVCaQjz7759lxryR0bD1vSUKme-ia_DIfJxy6BYfp0vURT3eU76irqw-A-DTdmfbY$>, or unsubscribehttps://urldefense.com/v3/__https://github.com/notifications/unsubscribe-auth/ADFBVB2KJVT62WUNJG63ZWDZXM2ATAVCNFSM6AAAAABNM3TFKKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGNRSGE2DINZWGE__;!!K-Hz7m0Vt54!m7AC66lt4XreJuPvgk-yhnYVCaQjz7759lxryR0bD1vSUKme-ia_DIfJxy6BYfp0vURT3eU76irqw-A-8gzBsPo$. You are receiving this because you were mentioned.Message ID: **@.***>

It seems odd to me too, but I followed the established pattern that already existed in Wikidata without questioning it. I've been ruffling enough feathers lately with questioning data modeling choices surrounding gender as it is. If someone else wanted to make some adjustments, though...

[CECSpecialistI]

I added another property to two of these to see if this seems appropriate for all of them. See https://www.wikidata.org/wiki/Q130324885 and https://www.wikidata.org/wiki/Q130324891https://www.wikidata.org/wiki/Q130324891. In each I added the property "code": [cid:810f38b9-4c0b-4e68-94da-21192a81fe09] [cid:b04fc9c6-0fba-4682-9a57-243315dd4b0b] Adam L. Schiff Principal Cataloger University of Washington Libraries (206) 543-8409 @. … ____ From: Crystal Yragui @.> Sent: Thursday, September 19, 2024 1:48 PM To: uwlib-cams/MARC2RDA @.> Cc: Adam L Schiff @.>; Mention @.> Subject: Re: [uwlib-cams/MARC2RDA] Produce IRIs for classification scheme datatypes (Issue #465) IRIs for classification scheme datatypes have been produced: https://w.wiki/BFBwhttps://urldefense.com/v3/__https://w.wiki/BFBw__;!!K-Hz7m0Vt54!m7AC66lt4XreJuPvgk-yhnYVCaQjz7759lxryR0bD1vSUKme-ia_DIfJxy6BYfp0vURT3eU76irqw-A-on_ZE_A$ — Reply to this email directly, view it on GitHub<https://urldefense.com/v3/__https://github.com/uwlib-cams/MARC2RDA/issues/465issuecomment-2362144761__;Iw!!K-Hz7m0Vt54!m7AC66lt4XreJuPvgk-yhnYVCaQjz7759lxryR0bD1vSUKme-ia_DIfJxy6BYfp0vURT3eU76irqw-A-DTdmfbY$>, or unsubscribehttps://urldefense.com/v3/__https://github.com/notifications/unsubscribe-auth/ADFBVB2KJVT62WUNJG63ZWDZXM2ATAVCNFSM6AAAAABNM3TFKKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGNRSGE2DINZWGE__;!!K-Hz7m0Vt54!m7AC66lt4XreJuPvgk-yhnYVCaQjz7759lxryR0bD1vSUKme-ia_DIfJxy6BYfp0vURT3eU76irqw-A-8gzBsPo$. You are receiving this because you were mentioned.Message ID: **@.***>

It doesn't look incorrect to me, do you imagine it would be used in a Wikidata to MARC scenario? It would be easy enough to generate a quickstatements sheet and upload similar statements to the rest of them, would probably take about 20 minutes.

[CECSpecialistI]

I could also very easily remove either the "part of" or "subclass of" statements depending on which one we wanted to go with, if we could be sure the Wikidata administrators wouldn't get huffy about it.

[CECSpecialistI]

But it's probably good to be consistent with the two pre-existing rdf:datatype items is what I was thinking.

[AdamSchiff]

I don't think an entity can be both an instance of and a part of the same thing. The semantics of the properties are different:

instance of: that class of which this subject is a particular example and member; different from P279 (subclass of); for example: K2 is an instance of mountain; volcano is a subclass of mountain (and an instance of volcanic landform)

part of: object of which the subject is a part (if this subject is already part of object A which is a part of object B, then please only make the subject part of object A), inverse property of "has part" (P527, see also "has parts of the class" (P2670))

subclass of: this item is a subclass (subset) of that item; all instances of this item are instances of that item; different from P31 (instance of), e.g.: volcano is a subclass of mountain (and an instance of volcanic landform); K2 is an instance of mountain

More explanation is at https://www.wikidata.org/wiki/Help:Basic_membership_properties

instance of (P31)

The relation between instances with a common feature and a class characterized by this feature is produced with the property instance of (P31). We use instance of (P31) instead of subclass of (P279) when we cannot say anything about instances with such relation. More specifically, it is an rdf:type.[1]

For example, Atlantic Ocean (Q97) and Pacific Ocean (Q98) are both instances of ocean (Q9430). We write therefore on Wikidata:

[Atlantic Ocean (Q97)](https://www.wikidata.org/wiki/Q97) [instance of (P31)](https://www.wikidata.org/wiki/Property:P31) [ocean (Q9430)](https://www.wikidata.org/wiki/Q9430);
[Pacific Ocean (Q98)](https://www.wikidata.org/wiki/Q98) [instance of (P31)](https://www.wikidata.org/wiki/Property:P31) [ocean (Q9430)](https://www.wikidata.org/wiki/Q9430).

subclass of (P279)

Is used to state that all the instances of one class are instances of another[[2]](https://www.wikidata.org/wiki/Help:Basic_membership_properties#cite_note-2)
More specifically, it is an rdfs:subClassOf[[3]](https://www.wikidata.org/wiki/Help:Basic_membership_properties#cite_note-3)‎
If any instances of the subject class are not instances of the value class, [subclass of (P279)](https://www.wikidata.org/wiki/Property:P279) is not appropriate. [partially coincident with (P1382)](https://www.wikidata.org/wiki/Property:P1382) may be used to link overlapping classes where neither is a strict subclass of the other.

Several examples:

[human brain (Q492038)](https://www.wikidata.org/wiki/Q492038) [subclass of (P279)](https://www.wikidata.org/wiki/Property:P279) [brain (Q1073)](https://www.wikidata.org/wiki/Q1073)
[brain (Q1073)](https://www.wikidata.org/wiki/Q1073) [subclass of (P279)](https://www.wikidata.org/wiki/Property:P279) [animal organ (Q24060765)](https://www.wikidata.org/wiki/Q24060765)
[animal organ (Q24060765)](https://www.wikidata.org/wiki/Q24060765) [subclass of (P279)](https://www.wikidata.org/wiki/Property:P279) [organ (Q712378)](https://www.wikidata.org/wiki/Q712378)

Another example:

class [lake (Q23397)](https://www.wikidata.org/wiki/Q23397) with instances [Lake Baikal (Q5513)](https://www.wikidata.org/wiki/Q5513) and [Lake Erie (Q5492)](https://www.wikidata.org/wiki/Q5492);
class [ocean (Q9430)](https://www.wikidata.org/wiki/Q9430) with instances [Atlantic Ocean (Q97)](https://www.wikidata.org/wiki/Q97) and [Pacific Ocean (Q98)](https://www.wikidata.org/wiki/Q98).

part of (P361)

Items cannot be an instance of another non-class, but they can be part of another non-class. For example, Albert Einstein's brain (Q2464312) is a part of Albert Einstein (Q937). On Wikidata we use part of (P361) for this relation.

[Albert Einstein's brain (Q2464312)](https://www.wikidata.org/wiki/Q2464312) [part of (P361)](https://www.wikidata.org/wiki/Property:P361) [Albert Einstein (Q937)](https://www.wikidata.org/wiki/Q937).

Classes can also be part of another class. For example,

[human brain (Q492038)](https://www.wikidata.org/wiki/Q492038) [part of (P361)](https://www.wikidata.org/wiki/Property:P361) [Homo sapiens (Q15978631)](https://www.wikidata.org/wiki/Q15978631).

The use of instance of (P31) instead of part of (P361) would be wrong here because human brain (Q492038) is not a person. Using subclass of (P279) would also be wrong because an instance of human brain (Q492038) (e.g. Albert Einstein's brain (Q2464312)) is not an instance of the class Homo sapiens (Q15978631).

Like subclass of (P279), part of (P361) is a transitive property.

There's also a table of examples showing the differences between these three properties. I do find it difficult to understand them, sometimes.

[CECSpecialistI]

I agree. I think it's an instance of (P31), personally. What do you think?

[AdamSchiff]

agree

Adam L. Schiff Principal Cataloger University of Washington Libraries (206) 543-8409 @.***

---

From: Crystal Yragui @.> Sent: Tuesday, September 24, 2024 5:45 PM To: uwlib-cams/MARC2RDA @.> Cc: Adam L Schiff @.>; Mention @.> Subject: Re: [uwlib-cams/MARC2RDA] Produce IRIs for classification scheme datatypes (Issue #465)

I agree. I think it's an instance of (P31), personally. What do you think?

— Reply to this email directly, view it on GitHubhttps://urldefense.com/v3/__https://github.com/uwlib-cams/MARC2RDA/issues/465*issuecomment-2372640181__;Iw!!K-Hz7m0Vt54!jtu3IErCOY3d6_nQ_2Xlj01NxgWYMUC5r3I8qkGFdp0U1Cq9NvawXV8SZVsCPqzf9cR6VGBLkRaqZXEYyTikrFk$, or unsubscribehttps://urldefense.com/v3/__https://github.com/notifications/unsubscribe-auth/ADFBVBYDIO5XFQ7WSBASZCLZYIBTPAVCNFSM6AAAAABNM3TFKKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGNZSGY2DAMJYGE__;!!K-Hz7m0Vt54!jtu3IErCOY3d6_nQ_2Xlj01NxgWYMUC5r3I8qkGFdp0U1Cq9NvawXV8SZVsCPqzf9cR6VGBLkRaqZXEYoyIEjgk$. You are receiving this because you were mentioned.Message ID: @.***>

[cspayne]

We need datatypes for Genre/Form Schemes