New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

NPGObjConXrefs: map RoleId and Prefix #49

Closed
VladimirAlexiev opened this Issue Jul 14, 2016 · 20 comments

Comments

Projects
None yet
6 participants
@VladimirAlexiev

VladimirAlexiev commented Jul 14, 2016

Currently the mapping maps only Prefix, and then only 2 values from it: 'Artist:' and 'Sitter:' (see NPGObjConXrefs-report.md). This takes care of 111020+125936=236956 rows, or 90.1% of the total 262833.
But it leaves a long tail of 10% of relations unmapped, i.e. just skips them. I made a pivot and shared it here: https://drive.google.com/folderview?id=0B7BFygWDV2_PYS0xMVR0V3pxVXc&usp=sharing:

image

  • Explain what the different RoleId mean (1 19 102 106)
  • Decide appropriate mappings for the top rows. Eg if we map the top 21, that will map 25.3k more rows, bringing the coverage to 99%.

Eg "Copy after" is very valuable art research info.

Related to #36 and possibly #37

@steads

This comment has been minimized.

steads commented Jul 14, 2016

Nice!
All the "possible" etc should be mapped assuming it to be true and then using CRMinf to capture the level of belief in the statement.

@azaroth42

This comment has been minimized.

azaroth42 commented Jul 29, 2016

Can we please avoid using CRMinf as one level of complexity too many?

@VladimirAlexiev

This comment has been minimized.

VladimirAlexiev commented Aug 1, 2016

Another option would be to use PX_likelihood as used by BM/ResearchSpace. It's a subprop of P2_has_type, applied on EX_Association, which is a subclass of E13

@steads

This comment has been minimized.

steads commented Aug 1, 2016

CRMinf describes the real complexity of argumentation. If you do not want this complexity you are free to use local solutions but they will be just that "local".

@si-npg

This comment has been minimized.

Collaborator

si-npg commented Aug 2, 2016

As with issue #39, many of these issues do not exist with our updated data. the Prefix field is populated for every ConXref in our updated data.
The four role IDs are as follows: 1=Artist, 19=Sitter, 102=Secondary Artist/Maker and 4=Associated Person/Institution. Sorry I did not submit our Roles table to clarify that.
Prefix, when it is not the same as the Role, is a more specific descriptor, like a sub-role.

@VladimirAlexiev

This comment has been minimized.

VladimirAlexiev commented Aug 9, 2016

@si-npg:

  • There are 4 values Suffix=" " that should be treated as (or replaced with NULL).
  • what is Role=106? Guess "Associated", seems 106 is now used instead of 4
  • Could you please explain what eg this situation means:
object=O, constituent=C, prefix=Copy after:, suffix=(Photographer)
  • could it be that "Cast after:" and "Copy after:(Sculptor)" mean the same?
  • should Role be ignored in the mapping (as it currently is)? If not, please explain it; eg "Publisher:" is used with 1, 102, 106)

Here is the updated pivot:
image

@si-npg

This comment has been minimized.

Collaborator

si-npg commented Aug 9, 2016

ROLES: 1=Artist, 19=Sitter, 102=Secondary Artist/Maker and 106=Associated Person/Institution
Most objects have a Sitter OR Associated Person/Institution; most have at least one Artist, and some have Secondary Artists/Makers too--often multiples.
Cast after: is used for a later cast after an original sculpture; Copy after: (Sculptor) is a copy by one sculptor--an actual artist--after another.
Prefix is not a controlled field. We do our best to keep it clean/consistent, but it's not perfect. It's used to give further explanation to the very broad Role term. For human-readability, it can always be used instead of Role, but we do rely very heavily on the Rold field itself.

@edgartdata

This comment has been minimized.

edgartdata commented Aug 9, 2016

@si-npg Great explanation of the implicit conventions hidden in plain language with:
Cast after: is used for a later cast after an original sculpture; Copy after: (Sculptor) is a copy by one sculptor--an actual artist--after another.
These would be modeled differently with the latter introducing the notion of influence from the original artist onto the artist who is copying the sculpture. I guess "cast after" could use P130_shows_features_of the original sculpture to the new cast?

@VladimirAlexiev

This comment has been minimized.

VladimirAlexiev commented Aug 10, 2016

@si-npg What does "object=O, constituent=C, prefix=Copy after:, suffix=(Photographer)" mean?
Does it mean that C created a photograph (not tracked in your system) that is a copy after O?
Or is O the identifier of the new object (the photograph), and then how is the old object identified?

About roles: I'm puzzled why the same Prefix is sometimes used with several Roles.

  • "Publisher" is most often used with role 102 "Secondary maker" and that makes sense.
  • But what does it mean "Publisher" with role 1 "Artist"? Is it different from the previous bullet?
  • And what does it mean "Publisher" with role 106 "Associated"?

@edgartdata: "Copy after: (Sculptor)" will map to crm:P130_shows_features_of. "Cast after" will map to more than that.
Here's a proposed mapping, I call these "CRM Casing". We have plenty of examples like this for CONA associative (artwork) relations:

  • "Copy after: (Sculptor)"
<newSculpture> crm:P130_shows_features_of <oldScuplture>.

<newSculpture> crm:P108i_was_produced_by <newSculpture-production>.
<newSculpture-production> crm:P14_carried_out_by <sculptor>.
  • "Cast after"
<newSculpture> crm:P130_shows_features_of <oldScuplture>.

<oldSculpture> crm:P108i_was_produced_by <oldSculpture-production>.
<oldSculpture-production> crm:P16_used_specific_object <oldSculpture-mold>.
<oldSculpture-mold> a crm:E22_Man-Made_Object;
  crm:P2_has_type aat:300024814. # mold (shaping equipment), aka cast (mold shaping equipment)

<newSculpture> crm:P108i_was_produced_by <newSculpture-production>.
<newSculpture-production> crm:P14_carried_out_by <sculptor>;
  crm:P16_used_specific_object <oldSculpture-mold>.
@VladimirAlexiev

This comment has been minimized.

VladimirAlexiev commented Aug 10, 2016

@azaroth42 @steads @workergnome @edgartdata @si-npg
Figure below made from above turtle using my rdfpuml tool:

image

As I said at the LA meeting in April, I think we need to make such prototypical models for all data areas. Otherwise I'm afraid confusion will reign: we can't be discussing this museum per museum, and in github issues. We need to document an overall model.

@edgartdata

This comment has been minimized.

edgartdata commented Aug 10, 2016

@VladimirAlexiev I am not sure why we would not want to use P_15_was_influenced_by for Copy after: (Sculptor) as that qualifier clearly emphasizes the influence of one artist on another?

@steads

This comment has been minimized.

steads commented Aug 10, 2016

"Copy after: (Sculptor)"

crm:P130_shows_features_of .

crm:P108i_was_produced_by .
crm:P14_carried_out_by .

This is not correct. The Copy after: (Sculptor) should be modelled with P15. This is showing the influence of a body of work or style not the copying of a particular object which is what the P130 shows.

@steads

This comment has been minimized.

steads commented Aug 10, 2016

Spot-on @edgartdata !

@VladimirAlexiev

This comment has been minimized.

VladimirAlexiev commented Aug 11, 2016

@si-npg states "Copy after: (Sculptor) is a copy", that's why I think P130 is relevant.

I agree that P15 should also be used: but it is that influenced the creation of , not the author of (only part of his work may have exerted that influence, and it could be it had several authors but only one of them had influence). The data only gives us evidence about , not other work(s) of the same author(s).

I'm puzzled that each row has only one objectID, @si-npg please answer my question above

Here's updated modeling of "cast after" including P15, how do you like it?
image

@si-npg

This comment has been minimized.

Collaborator

si-npg commented Aug 15, 2016

Q: What does "object=O, constituent=C, prefix=Copy after:, suffix=(Photographer)" mean?
Does it mean that C created a photograph (not tracked in your system) that is a copy after O?
Or is O the identifier of the new object (the photograph), and then how is the old object identified?

A: O refers to the new object, which has a different medium than the original (which is why we add a suffix in this case; normally a copy is a copy in the same medium.) So the example above describes a an object in some medium other than photography that is a copy after a photograph.

Q: About roles: I'm puzzled why the same Prefix is sometimes used with several Roles.
"Publisher" is most often used with role 102 "Secondary maker" and that makes sense.
But what does it mean "Publisher" with role 1 "Artist"? Is it different from the previous bullet?
And what does it mean "Publisher" with role 106 "Associated"?

A: Publisher in the role of artist occurs in 4 NPG objects, and I have requested that our curators to change these. (They were entered this way intenionally, but it's not necessary.) Publisher in the role of Associated Person occurrs in just 2 NPG objects, and I've already changed these, since they were just mistakes! So the Publisher prefix will not be an issue going forward. However, there may be other prifixes that are sometime used for one role and sometimes another. (I'll need to check whether any of these are included in our data set.)

@VladimirAlexiev

This comment has been minimized.

VladimirAlexiev commented Aug 16, 2016

@si-npg Thanks for the answers!
Just for clarity: so the old object and its creator are not mentioned in this table?

@si-npg

This comment has been minimized.

Collaborator

si-npg commented Aug 16, 2016

The artist of the original object is entered as a "Secondary Artist" and does have a link to the object via the ConXrefs table.
The original object itself, if it is entered in the database, is linked to the copy, but those object-to-object links are not included in our exported data. Additional relevant fields that I did not export are Historical Dates, where the date of the original is entered for objects that are copies; and Related Works and/or Notes--text fields which might reference the original work.

@VladimirAlexiev

This comment has been minimized.

VladimirAlexiev commented Aug 16, 2016

  • so I guess the two correlated records would have:
"Artist:" <new artist>, "Copy after:" <old artist>
  • but how to interpret a suffix? Should it be like this?
"Artist:" <new artist> "(Photographer)", "Copy after:" <old artist>
"Artist:" <new artist> "(Sculptor)", "Copy after:" <old artist>
  • or like that
"Artist:" <new artist>, "Copy after:" <old artist> "(Photographer)"
"Artist:" <new artist>, "Copy after:" <old artist> "(Sculptor)"

Sorry to dig like this, but @azaroth42 we need to grok a similar situation for JPMG.

@si-npg

This comment has been minimized.

Collaborator

si-npg commented Aug 16, 2016

Your second interpretation is right. We use the suffix to describe the artist of the old (original) work.

@VladimirAlexiev

This comment has been minimized.

VladimirAlexiev commented Aug 22, 2016

Ok. I also notice that

  • Copy After goes mostly with role=102 (the old master only influenced the new work)
  • but Cast After goes mostly with role=1 (the new master is a prime contributor to the new work)

Let's consider the following situation:

  • npg/constituent/1 made copies of two works by older masters:
  • he made an artistic copy npg/object/1 of a work by npg/constituent/2
  • he made a literal copy npg/object/2 of a work by npg/constituent/3 from a mold (cast)
ConXrefID ObjectID ConstituentID RoleID Prefix Suffix
1 1 1 1 Artist:
2 1 2 102 Copy after: (sculptor)
3 2 1 1 Artist:
4 2 3 1 Cast after:

(Remember we don't have links to these older works)

It could be modeled like this. I don't quite like that it uses completely different constructs for the two cases. We could also make a P15 link in the second case.
@si-npg @azaroth42 @steads feedback is welcome!

<npg/object/1> crm:P108i_was_produced_by <npg/object/1-production>.
<npg/object/1-production> a crm:E12_Production;
  crm:P14_carried_out_by <npg/constituent/1>;
  crm:P15_was_influenced_by <npg/constituent/2>.
<npg/object/1/contribution/2> a crm:E13_Attribute_Assignment;
  crm:P140_assigned_attribute_to <npg/object/1-production>;
  crmx:property crm:P15_was_influenced_by;
  crm:P141_assigned <npg/constituent/2>;
  crm:P2_has_type <thesaurus/contribution/copyAfter>;
  crm:P2_has_type <thesaurus/contribution/sculptor>. 

<npg/object/2> crm:P108i_was_produced_by <npg/object/2-production>.
<npg/object/2-production> a crm:E12_Production;
  crm:P14_carried_out_by <npg/constituent/1>;
  crm:P16_used_specific_object <npg/object/2-mold>.
<npg/object/2-mold> a crm:E22_Man-Made_Object;
  crm:P2_has_type aat:300024814;
  crm:P108i_was_produced_by <npg/object/2-mold-production>.
<npg/object/2-mold-production> a crm:E12_Production;
  crm:P14_carried_out_by <npg/constituent/3>.

<thesaurus/contribution/copyAfter> a puml:Inline.
<thesaurus/contribution/sculptor>  a puml:Inline.
aat:300024814 a puml:Inline; rdfs:label "mold (shaping equipment)".

image

@bsnikhila bsnikhila closed this Apr 30, 2017

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment