Issue 202: Explanation of "multiple instantiation" concept at the "Disjointness"

Starting Date: 
2012-01-04
Working Group: 
1
Status: 
Done
Closing Date: 
2013-10-23
Background: 

Posted by Michael Hopwood 20/12/2011 

For my second posting to this list I wanted to ask a related question to my previous one: has anyone to your knowledge tried to model commercial objects in CRM?

I know that the FRBR(oo) harmonisation introduced many useful distinctions relevant to published books, but, by analogy, the idea here is whether you could equally well model commercially released music (in CD format, or digital files), audiovisual (TV and features) in DVD format, streaming media or downloads, or commercially available images like stock photography, or "physical" prints of classic(al) images like historic photographs or photos of art works? 

The main problem to be addressed seems to be that CIDOC-CRM aims to describe unique objects, but commercial products are precisely designed to be replicable and interchangeable. I think FRBR(oo) addresses this but is anyone else interested in developing this aspect further? 

I wondered if there might be a possible "cross-over" for the cases of commercially available cultural heritage items such as books published for particular items or exhibitions/events by/in partnership with institutions, replicas of iconic artefacts, the obvious prints of photos of paintings, compilations of sound recordings from archives, educational DVDs? Does CRM already describe these adequately?
 



Posted by Anna Jordanous 4/1/2012 

My impression of FRBRoo is that the FRBRoo harmonisation is applicable to any Work which is published, particularly in publication runs. This is not just limited to Linguistic Objects, as shown by the documentation and examples for FRBRoo for e.g. F24_Publication_Expression, F26_Recording. So FRBRoo should be appropriate for your purposes? 

I'm looking into FRBRoo to model medieval manuscripts which are replicated through (edited) copying by scribes. In fact I have a problem related to modelling published written works with FRBRoo. Hope you don't mind if I borrow your thread to ask the list about this: 

I would like to model information about physically existing manuscripts such as the Language a manuscript is written in, and record relationships between manuscripts such as if one manuscript is a translation of another. Therefore, I would like to treat a frbroo:Manifestation_Singleton (subclass of crm:Physical_Man-Made_Thing) as a crm:Linguistic_Object, which is down the crm:Conceptual_Object branch. However, ideally I want to avoid making a subclass of crm:Physical_Man-Made_Thing that is also a subclass of crm:Conceptual_Object (I know these two classes are not disjoint, but this union seems an unnatural one to make). 

The preferred approach, I assume, would be to abstract to the Expression represented in the Manifestation_Singleton, and then record language/translation information, but this seems somewhat longwinded. 

So essentially I am asking: is there a better way in CIDOC/FRBRoo to represent a physically existing Linguistic Object? Would welcome any advice or thoughts on this?
 



Posted by Joao Lima 4/1/2012 

According to CIDOC CRM (v. 5.0.4, p. xvi), "E18 Physical Thing is disjoint from E28 Conceptual Object". So, IMHO, you couldn't subsume Linguistic_Object to Manifestation_Singleton.

However, the CIDOC CRM allows multiple instantiation, ie, one specific manuscript could be, at the same time, instance of Manifestation_Singleton and Linguistic_Object classes. See at FRBRoo (v. 1.0.1, p. 14, Fig. 1) that the "F28 Expression Creation" event produces (simultaneously) an Expression and a Manifestation Singleton instance. The same event could also create the "Linguistic Object" instance.
 


Posted by Patrick Le Boeuf 4/1/2012 

Dear all, 
(1) To answer Michael Hopwood:

Yes, it is quite possible to use a combination of FRBRoo and CIDOC CRM to model commercially available reproductions of unique objects. Possible paths include: 

a) "replicas of iconic artefacts:"
E22 Man-Made Object [= the reproduced unique artefact] P16B was used for (P16.1 mode of use: E55 Type {source for reproduction}) F30 Publication Event
F30 Publication Event R24 createdF24 Publication Expression [= the set of signs present on the commercial product, including its packaging]
F24 Publication Expression CLR6B should be carried by F3 Manifestation Product Type [= the commercial product]
F24 Publication Expression P130 shows features of (P130.1 kind of similarity: E55 Type {commercialized replica}) E22 Man-Made Object [= the reproduced unique artefact]
F24 Publication Expression R27B was used by F32 Carrier Production Event [= the industrial process through which all individual exemplars of the product are made]
F32 Carrier Production Event R28 produced F5 Item [= each individual physical exemplar of the commercial product]

b) "prints of photos of paintings:"
E22 Man-Made Object [= the photographed painting] P16B was used for (P16.1 mode of use: E55 Type {photographed item}) F29 Recording Event 
F29 Recording Event P2 has typeE55 Type {making photographs} 
F29 Recording Event R21 createdF26 Recording [= the set of signs present on the photograph of the painting that was used as source for the publication] 
R26 Recording P2 has type E55 Type {photograph} 
F26 Recording R14B is incorporated in F24 Publication Expression [= the set of signs present on the commercial product, including its packaging] 
F24 Publication Expression CLR6B should be carried by F3 Manifestation Product Type [= the commercial product] 
F24 Publication Expression P130 shows features of (P130.1 kind of similarity: E55 Type {commercialized photograph}) E22 Man-Made Object [= the photographed painting] 
F24 Publication Expression R27B was used by F32 Carrier Production Event [= the industrial process through which all individual exemplars of the product are made] 
F32 Carrier Production Event R28 produced F5 Item [= each individual physical exemplar of the commercial product] 

c) "compilations of sound recordings from archives:" 
F26 Recording [= the content of sound archives] R14B is incorporated in F24 Publication Expression [= the set of signs present on the commercial product, including its packaging] 
F24 Publication Expression R27B was used by F32 Carrier Production Event [= the industrial process through which all individual exemplars of the product are made] 
F32 Carrier Production Event R28 produced F5 Item [= each individual physical exemplar of the commercial product] 

d) Exhibition catalogues and educational DVDs are modelled exactly the same way as any book and any DVD, see FRBRoo. 

(2) To answer Anna Jordanous:

The only way to connect a physical manuscript with a language is to go through its conceptual content. If you use CIDOC CRM only, this can be done as follows: E84 Information Carrier P128 carries E33 Linguistic Object P72 has language E56 Language. If you use a combination of CIDOC CRM and FRBRoo, you can use this path: F4 Manifestation Singleton P128 carries E33 Linguistic Object P72 has language E56 Language. In CIDOC CRM, E18 Physical Thing is explicitly declared as disjoint from E28 Conceptual Thing (it is one of the very few cases of explicit disjointness in CIDOC CRM), it is therefore strictly forbidden to regard F4 Manifestation Singleton as a subclass of E33 Linguistic Object. It is equally impossible to use multiple instantiation to declare one thing as simultaneously an instance of F4 Manifestation Singleton and of E33 Linguistic Object, as Joao Lima suggests, because the definition of two disjoint classes is precisely that an instance of one of the two classes can never be simultanesouly instantiated as an instance of the other class.

This is not just to bother you. Actually, this is perfectly logical: it does not make sense (although it is everyday parlance) to say that "a manuscript is a translation of another." A manuscript is basically just a physical conglomerate of parchment (or paper), ink, and binding materials, it can't be the "translation" of anything. What is a translation of something is the conceptual content referred to by the particular arrangement of the ink on the parchment (or paper). If you were happy enough to discover a manuscript that would be "a translation of" the Voynich Manuscript (without its illustrations), you would even not notice it, because no one has access to the conceptual content of the Voynich Manuscript, and no one is therefore in a position to compare it with its translation.
 



Posted by Joao Lima 

Thanks for the correction of my improper use of "multiple instantiation" concept. In my example, the "multiple instantiation" concept could only be applied to the "F2 Expression" and "E33 Linguistic Object" instances, is that right? 
In addition, I should remember that the "E33 Linguistic Object" class has the "P73 has translation" property that could be used to connect the two entities. 

PS. Maybe the CIDOC CRM document could explain the "multiple instantiation" concept at the "Disjointness" section (CIDOC, v. 5.0,4, p.xvi) 
 



Posted by Martin 4/01/2012 

I take your proposal as an issue to be treated in the next meeting: "PS. Maybe the CIDOC CRM document could explain the "multiple instantiation" concept at the "Disjointness" section (CIDOC, v. 5.0,4, p.xvi)." 

Someone volunteering to write a text?

Current Proposal: 

Put a reference to multiple instantiation under "disjointness", add paragraph to terminology. 

Martin will write a note 
Heraklion Crete, 3/5/2012 
 


Posted by Martin 30/9/2013 

"Multiple Instantiation: 
Multiple Instantiation means that an instance of class A is also declared as an instance of one or more other classes B1...n at the same time. For instance, some particular cases of destruction may also be activities, but not all destructions are also activities. Whereas inheritance has the effect that any instance of a class A must implicitly be instance of all superclasses of A by virtue of the definition of the class A, the combination of classes used for multiple instantiation is a characteristic of particular instances only. Multiple instantiation is not allowed using combinations of disjoint classes. "

Outcome: 

The text accepted by the CRM-SIG is the following: 

"Multiple Instantiation is the term that describes the case that an instance of class A is also regarded as an instance of one or more other classes B1...n at the same time. When multiple instantiation is used, it has the effect that the properties of all these classes become available to describe this instance. For instance, some particular cases of destruction may also be activities(e.g.,Herostratos' deed), but not all destructions are activities (e.g., destruction of Herculaneum). In comparison, multiple inheritance describes the case that all instances of a class A are implicitly instances of all superclasses of A, by virtue of the definition of the class A, whereas the combination of classes used for multiple instantiation is a characteristic of particular instances only. It is important to note that multiple instantiation is not allowed using combinations of disjoint classes." 

29th CRM-SIG meeting, Heraklion, October 2013