Issue 302: Examples of A6 Group Declaration Event, A7 Embedding, A8 Stratigraphic Unit

Starting Date: 
2016-07-29
Working Group: 
3
Status: 
Open
Background: 

The example in the  entities A6 Group Declaration Event, A7 Embedding, A8 Stratigraphic Unit is the same. The problem is that the same example is used  for describing an event, a state and a feature

Lida Charami, Heraklio 29/7/2016

Current Proposal: 

In the 36th joined meeting of the CIDOC CRM SIG and ISO/TC46/SC4/WG9 and the 29th FRBR - CIDOC CRM Harmonization meeting, the sig decided for:

(a) A6 Group Declaration Event: to keep the example and in order to take into account other type of grouping events, this class might have to be broadened. PIN will take on changing scope note of A6 and broaden the range of AP16 assigned attribute to (was attributed by): A8 Stratigraphic Unit ( E18?). Also they should consider to specialize P141 property  giving  new  labels (declared’ or ‘declared as remains’) and consult with CRM survey team (this will  be facilitated by GB) 
(b) A7 Embedding, A8 Stratigraphic Unit: the examples will be changed by Pin

Heraklion, 1/8/2016

has been sent  by George to Tymon de Haas and Martijn van Leusen on 20/10/2016

I hope this mail finds you well. I am writing because at our last CRM SIG meeting a discussion came up with regards to the CRMArchaeo class A6 Group Declaration Event. This class was designed for documenting activities in which some stratigraphic units are attributed some common attribute. As it is currently formulated it is limited to this act. But the examples from your work suggest that such a definition might be too restrictive and the class could be broadened in intention to cover more acts of grouping objects due to an act of reasoning which sees common characteristics in some assemblage of archaeological objects.

Here is the original document: http://www.ics.forth.gr/isl/CRMext/CRMarchaeo/docs/CRMarchaeo_v1.4.pdf

The proposal is that likely we would want to be able to say of any Physical Thing that it can be attributed some common attribute and thus become a group. The PIN laboratory of which Achille Felicetti is a scientific member maintains this standard. I thought I would write to solicit your opinion and open the conversation, should you have time.


has been sent by Tymon de Hass on 24/10/2016

thank you for this. At first sight, I am not sure if this class would actually be appropriate for the types of attribute assignment we are dealing with: in the case of artefact classification, we are subdividing a group of objects into subgroups - these subgroups do have sets of common characteristics, but are not first defined as single physical objects and then attributed a common attribute (as in the case of the postholes). I am thinking whether there are situations in which this is applicable - is the attribution of diagnostic artefacts to some typological form perhaps a similar act? (eg, saying sherds x, y and z because of their morphological characteristics can all be interpreted as belonging to a coarse ware jar of type AA)?


 has been sent by  Martijn van Leusen on 24/10/2016

At first sight any type of classification act can be described as a group declaration event. However, these groups can exist without any members (eg, the class 'black gloss ware' may be empty for a given site or survey unit), which is very different from declaring three actual sherds to be part of this group. Also, the group formed thusly does not need to have a 'genetic' relationship such as a set of postholes belonging to the same structure would have. Or am I overthinking this? Maybe the salient point is simply that somebody 'declares' that one or more things belong to a 'group', and the details are described in a DorP ? And the group could be (eg) of type 'ware'?

has been sent by Achille on 11/11/2016

Sorry if I join so late this very interesting discussion. Just a quick note to clarify.
 
I have the impression that here we are talking about slightly different things. When Martijn speaks of “empty groups without members” and Tymon of “artefacts classification”, I have the impression that they probably have in mind the assignment of one or more objects to a given class or type, namely the classification of real objects with respect to an abstract class, something very similar to the E17 Type Assignment event of the CIDOC CRM core. We must try not to confuse this kind of activities with the one provided by the A6 Group Declaration Event in CRMarchaeo, whose function is somewhat different. In the case of AP6, in fact, the event is to recognise two or more real objects or stratigraphic units as part of a single real object or stratigraphic unit, regardless of their type. This could happen, for example, when archaeologists realise that two portions of wall belong to the same building or five terracotta fragments belong to the same vessel, regardless of the type of the building or the vessel itself. Certainly the vessel and the wall can then be assigned to one or more typologies, as Martijn pointed; but this is another event.

I agree with George that, as it currently defined, the class is not yet fully able to describe this event in all its nuance because it is limited to the stratigraphic units only. But we are going to modify its definition to extend its scope (and the range of the corresponding AP16 property) in order to comprise every physical object involved in an archaeological excavation, not only the stratigraphic units.
 
The new scope note will be something like this:
 
"A6 Group Declaration Event: This class comprises activities resulting in the assignment of a common attribute to several Stratigraphic Units (A8), Stratigraphic Building Units (B5) and Physical Man-Made Thing (E18). This may be due to an archaeological interpretation of them being part of one physical thing, like postholes being part of one building, fragmentary structures being part of a wall or pieces of clay being part of an artefact.”
 
This should be enough to guarantee flexibility to this kind of event. But of course, every suggestion or comment is welcome.

has been sent by Martin on 11/11/2016

Dear All,

My apologies that I didn't have the time to follow the arguments, I hope I can catch up with George soon.
Only one substantial remark:

I absolutely agree with Achille. The distinction between particulars and universals must not be confused at any time. I'd prefer the Grouping Event excusively to be used for declaring a whole, or any particular constellation. Classification of artifacts is a process of "induction", an attempt of a universal theory. The whole process is completely modelled in the CRM in analogy to biology, together with the description of the evidence, which always starts with an aggregate of samples. If we need something like "potential types", or "type assumption", this should fall under and extend the "taxonomic discourse". We must not introduce competing classes in the model. If the distinction between particular states of affairs and universal may turn out ambiguous in practice in some rare cases, we need to analyze this very carefully and take according measures/create rules. I believe a Group of particulars is always a set of complementary elements, such as 4 wheels making up a cart, but never five. Even if a set of equal coins is found, its being a whole is due to a deposition event (someone dropped it, left a purse in a burning building, hid it), and NOT due to the type. It is complementary to the event.
If the set is grouped because of the type, or to recgnize a type, and not because of the reason for being together, it is a classification process.

Would that make sense?

If yes, I think the current scope note is bad: "This class comprises activities resulting in the assignment of a common attribute" . This sounds exactly like classification....


has been sent by Achille on 11/11/2016

True, the scope note of this class definitely needs a substantial revision. We will discuss and try to reformulate it shortly to make it consistent with the real intention of the class itself.
Thank you very much for your precious considerations on this topic. Any further suggestion will always be welcome to us.

has been sent by Martijn van Leusen on 14/11/2016

I thought Achille's view was clear and convincing, and conclude from it that the decision to put a specific sherd into the finds class 'Black gloss' is NOT a A6 Group Declaration Event. However, I am still a bit confused about the alternative. Do we now agree that this can (and should) be a E17 Type Assignment event? 

And if the answer is Yes, what is the proper way to deal with the plethora of types that archaeologists have invented over the centuries? Some of these (like 'black gloss') may already live in some vocabulary/thesaurus somewhere, but during the actual act of assignment we were certainly not using that specific list of types plus criteria, but our own (unfortunately underspecified) one. Should we point to our own specific DorP for the details of the type assignment?

has been sent by Achille on 15/11/2016

To reply to your questions I would like first of all to underline the fact that the assignment of a type may have different levels of detail in CIDOC CRM and the use of the E17 Type Assignment event is not always necessary, especially when what you need is just a plain statement, i.e., that an object, an actor, an event or any other entity in your domain, belongs to a given type. Usually the quickest way to do this is through the property “P2 has type”, for instance as in the following example:
 
E22 Man-Made Object -> P22 has type -> E55 Type (“Anfora”)

In most of the cases, this is already to describe the mere condition of the “belonging of an entity to a given type”, without any need to further specifications, e.g., on how and by whom this assignment was made, or on what date and in which place it happened. The E17 class should be used only in the presence of similar specific needs, when it becomes necessary to provide more details on the attribution itself as a specific event and to document the process of stating a subjective opinion that ended up with the assignment itself; in this case it is possible to build a statement like the following:
 
E22 Man-Made Object -> P41i was classified by -> E17 Type Assignment -> P42 assigned -> E55 Type (“Anfora”).
 
Being the E17 Type Assignment an event, as we said, additional relationships can be deployed in order to describe its space-temporal aspects and the actors and objects involved in it.
 
Please notice that the latter statement is not in contradiction with the former one; each of them just provides a different level of detail, a different description of the same thing, and can be chosen according with the specific knowledge (i.e., the amount of information) available in the domain you are describing.

Regarding your second question, I would underline that CIDOC CRM is not concerned in any way with, and is in any way supposed to provide or recommend the use of one vocabulary/thesaurus/list of terms rather than of another. The choice of the typologies to be used in a specific domain is entirely entrusted to the users, i.e., the experts of that specific domain. The choice of effective terminological tools for documenting archeology, for example, is a decision that rests solely with the archaeologists, being the creators of the ontology only responsible to provide users with the necessary tools for the conceptual (and not the terminological) description of their domain.

The term “Anfora” that I used in the previous example, could be arbitrarily chosen from any local/internal taxonomy, a widely adopted vocabulary or an international thesaurus, like, for example, the Getty’s Art & Architecture Thesaurus (AAT). In this case, the previous CIDOC CRM triple would look like this:
 
E22 Man-Made Object -> P22 has type -> E55 Type (“http://vocab.getty.edu/aat/300148696”)
 
with the “http://vocab.getty.edu/aat/300148696” URI of the AAT thesaurus used for representing in a more “Semantic Web” fashion the “amphorae” concept of AAT. All these statements will continue to be perfectly valid in terms of CIDOC CRM, regardless of the values assigned to the E55 class.
 
I hope this clarifies.

has been sent by Martijn van Leusen on 15/11/2016

that is very useful, thank you. Just to complete this thread, our current proposal for a survey CRM would end up having many lists of E55 Types: there would be pottery fabric types, ware types, feature sherd types, types of survey unit, etcetera. Where/how would you store information about each of these types (e.g. the criteria used to decide that something belongs to the E55 Type "coarse ware")? The point being that anybody using our databases should be able to check for themselves that our "coarse ware" is the same stuff as the the "coarse ware" in some other archaeologists' database....

Sorry if I sound like a newbie here!

has been sent by Martin on 15/11/2016

The best is you create a SKOS Scheme. For each basic class that is typed by your vocabulary, i.e., E7 Activity, Physical Object, Physical Feature, etc., you can create a different skos:Collection of terms (i.e., "object terms","activity terms"...). Best use "broaderTransitive" to descibe hierarchies of terms. As long as SKOS is used to describe universals, and not persons or particular places, skos:Concept is "same as" E55 Type. SKOS foresees all the descriptional properties. You can propose the vocabulary as standard, and you can map with SKOS other vocabularies to it. S/W can reason on SKOS hierarchies.

 

 

posted by Achille on 29/11/2016

Dear all,

Please find below a new tentative scope note for the CRMarchaeo “A6 Group Declaration Event” class:

========================================
A6 Group Declaration Event: This class comprises interpretive activities that lead to identify an originary/previous unity (singularity) of two or more Stratigraphic Units (A8), Stratigraphic Building Units (B5) and Physical Man-Made Thing (E18). The activity of archaeological interpretation consists of the recognition of them being part of one physical ‘thing’ or object, like two stratigraphic units (with no evident contact) cut through by a ditch being part of the same original stratigraphic unit, two or more surviving parts of a structure being part of the same wall or a number of potsherds being part of the same original artefact.
========================================

Comments and suggestions welcome. 

posted by Martin on 29/11/2016

Dear Achille,

Let me make the times associated more explicit:

A6 Group Declaration Event: This class comprises interpretive activities that lead to identify two or more instances of Stratigraphic Units (A8) and / or Physical Thing (E18) that simultaneously exist at the time of this activity or at the time of an archaeological observation this activity refers to as source and that are attributed to be the remains of one complete instance of Physical Thing (E18) that had existed at a time of reference in the past, such as two stratigraphic units (with no evident contact) cut through by a ditch having been segments of the same original stratigraphic unit, two or more surviving parts of a structure having been segments of the same wall (B5), a number of postholes being the indication of a past wooden house or a number of potsherds being segments of the same original artefact.

Comments:
Stratigraphic Building Units (B5) are subclass of A8, hence do not contribute to the definition, but could appear as additional comments.

I suggest not to speak of "parts" in case of remains of pieces of coherent matter as a result of destructive process. At the time we are interested in, the object had no parts. It was one. However, the structure at our hands formed a segment in the sense of Gerstl & Pribenow (i.e., a constellation of matter with no natural boundaries at the time the thing existed as a whole or at all, and no knowledge of such boundaries prior to the destruction).

I wonder if we should restrict to man-made objects. See attached image. Imagine the stone, without the stand, dispersed in the remains of a burned-down temple?

In first order logic, I'd put in a constraint that the STV of the target thing has to be "before" the activity. 

posted by Paola Ronzino on 30/11/2016

Dear Martin, 

Thank you very much for your proposal, which we totally agree with. 
 
We would like to point out that B5 is not a direct subclass of A8, but a subclass of A2 because it concerns only with matter. Does this affect the definition? 
Many thanks in advance for your clarification.
 

posted by Martin on 29/11/2016

Dear All,
after discussion with Paola Ronzino, Achille Felicetti, George Bruseker, Chrysoula Bekiari , Eleni Christaki

We propose:
 A6 Group Declaration Event:

This class comprises interpretive activities that lead to identify two or more instances of Stratigraphic Units (A8) and / or Physical Thing (E18) that simultaneously exist at the time of this activity or at the time of archaeological observations this activity refers to as source and that are attributed to be the remains of one complete instance of Physical Thing (E18) that had existed at a time of reference in the past, such as two stratigraphic units (with no evident contact) cut through by a ditch having been segments of the same original stratigraphic unit, two or more surviving parts of a structure having been segments of the same wall (B5), a number of postholes being the indication of a past wooden house or a number of potsherds being segments of the same original artefact.

Comments:

I suggest not to speak of "parts" in case of remains of pieces of coherent matter as a result of destructive process. At the time we are interested in, the object had no parts. It was one. However, the structure at our hands formed a segment in the sense of Gerstl & Pribenow (i.e., a constellation of matter with no natural boundaries at the time the thing existed as a whole or at all, and no knowledge of such boundaries prior to the destruction).

In first order logic, I'd put in a constraint that the STV of the target thing has to be "before" the activity. 

 

 

In the 37th joined meeting of the CIDOC CRM SIG and ISO/TC46/SC4/WG9 and the 30th   FRBR - CIDOC CRM Harmonization meetingthe crm-sig reviewed the proposal made by Martin about A6 Group Declaration Event and assigned to GB and Steve to check the syntax considering the follow comments

o At the time of reference of being destroyed and time of reference in the past   the two things shared complete identity.
o B5 should be removed. 
o To investigate if we should restrict to man-made objects.  (See Martin’s comment about the stone in the image and imagine the stone, without the stand, dispersed in the remains of a burned-down temple!)
o Think about a constraint in  first order logic   that the STV of the target thing has to be "before" the activity.


Then the crm – sig reviewed and made changes to the proposal made by Achilles about AP16. They decided that P141 cannot be used here since the range has been restricted. Need to declare a subclass and assigned to homework to GB to make the scope more comprehensible


The revision of the examples of A7 and A8 is still pending.

Berlin, December 2016