Modeling Gender

[chin-rcip]

I put this topic into discussion because I think we found a temporary solution, but we still need to think more about this issue and how to model such complex and sensitive matter.

If we agree that Nationality, Nationhood and Communities are groups that you belong to, with a joining and leaving event (either by birth or by citizenship process), things are different for cultural affiliation and gender.

• Cultural affiliation, understood as cultural milieu we live in and that we keep all our life, does not have a joining and leaving event. A Swiss person, born and raised his first years of his life in Switzerland and then moved out to Canada during his teenage years, would have the Swiss culture, but would also imbued the Canadian culture if he stays in Canada for some years.
  Therefore, the modeling of the Culture with the Group Belonging pattern is not correct, and maybe the P2_has_type ->E55_Type would be more accurate.

• Gender belonging is on the contrary much more complex than the P2_has_type->E55_Type pattern. Nonetheless, the Group Belonging is not appropriate as gender can not be modeled as a E74 Group.
  Linked.art and CIDOC CRMSoc are developing a "Phase" concept, that would allow to document those kind of "non-physical statements" about a E21 Person. It is better to wait and see how this is developed.
  Another solution, proposed by George Bruseker, would be to create a "declaration event", that would model the gender as a declaration statement, either made by the person or an institution. I am not really in favor of creating our own classes.

[Habennin]

The CRM SIG recommends making your own classes, aligned to CIDOC CRM, when CRM does not have the expressive capacity you require. Nevertheless if you wanted to use the 'declaration event' idea without creating a new class then you could simply use E13 Attribute Assignment where the individual doing the attributing and the thing being attributed are the same.

ie:

E13 -> p2 -> E55 "Gender Declaration"
E13 ->p 140 -> E21
E13 -> p141 -> E55 Type "Bisexual" -> p2 -> E55 Type "Gender"
E13-> p14 -> E21 "Joe"
E13 - p4 -> E52 "such and such a data 1982"

The phase class is still heavily debated even in Linked.art.

[eecanning]

The LGBTQ+ Linked Data Vocabulary Homosaurus - http://homosaurus.org/ - I think can help point to a solution, both in terms of an authoritative vocabulary source and in terms of borrowing from their structure. For example, when it comes to gender, Homosaurus recognizes distinct terms of "assigned gender" and "gender identity" (as well as "culturally specific gender identities"). Assigned-at-birth terms (ie. AFAB/AMAB) could be seen as the activity of the assignation and resulting identity (ie. "Gender Assignment at Birth") with something like @Habennin's "Gender Declaration" activity used to make space for the assertion of non-cis identities and changing identities. P127 could link the broader notion of E55 "Gender Identity" with "Assigned Gender" and "Declared Gender".

[Habennin]

That's a very useful resource, thanks! Also the idea of the 'assigned at birth gender' formulation makes sense. Then you make it clear that this is not what the person says of themselves, but only what an administrative system does? Is that a correct reiteration of what you suggest? So we could actually just one pattern but you would either say the person themselves does the attributing or so institutional source.

This would make good sense to me.

[stephenhart8]

Thank you for this, it is indeed very useful!
So if I'm following you both, we could have to options for gender identity.

1. The first on with the E55 Type, but make a difference between the gender assigned and the gender declared, as in the following pattern and example:
   
   In this example, actor/001 is a female, that was declared male at birth.

2. In the E13 Attribute Assignment I am not sure to be right, but I would see things like that:
   
   Is that what you had in mind @Habennin ?

[Habennin]

can you share the draw.io base diagram and I'll present my alternative. More or less we are on the same page.

[stephenhart8]

Sorry I forgot to put them.

1. https://drive.google.com/open?id=1bbgFuocOZXIVM4nYlOXd2ECtCFfe-L4o
2. https://drive.google.com/open?id=1SBc4hWLCJZ7SeiXfXQIyI6xGxCY9V7w_

[stephenhart8]

Let me know if you have access problems

[VladimirAlexiev]

I think your Swiss and Canadian examples are both Groups, it doesn't matter one was acquired by birth and another through residence. Using Group allows you to say an artwork was made by unknown Swiss or unknown Canadian; E55 Type is inferior for this.

(Come think of it, why should we exclude "nation by conviction"? If I really want to be Qatari -of the rich kind- who should judge me I can't call myself that?)

[stephenhart8]

@VladimirAlexiev Groups are indeed used for nationality. But as the Cultural Affiliation is more of a loose attribution, we decided to use another pattern.
For the Unknown artists, it seems better for us to have an artist that just does not have an appellation (we do not know his name) but is linked to this cultural affiliation or nationality.

@Habennin What are your alternatives to the modelization of the gender? I think the Phase would be a great option for it. Will documentation on it be released soon?

---

We've decided to separate this issue in two:

• This one will continue to focus on the gender pattern
• I've created issue Cultural Affiliation and its modelization #50 to discuss the Cultural Affiliation field and its modelization

[marielmat]

In preparation for the 11th meeting of the Semantics Committee (June 10).

Here is a recording to the presentation made about Homosaurus Vocabulary during the 2020 LD4 Conference : https://ld42020.sched.com/event/cjKm/workshopping-queeries-linked-data-vocabularies-and-ethical-cataloging.
And the link to the LD4 Ethics in link data group : https://drive.google.com/drive/folders/1X7ZBtSVTjOKJ7FUeRYk2NWdHDqsvg5BK?usp=sharing

We did not document gender for several years, but this issue comes up often when we want to analyze our data. For example, we can't provide data about the presence of women artists in our collections...

Having said that, the fact is that our terms are limited to "homme" and "femme" (gender identity) and that this attribution has often been made by the person who catalogs and on the basis of the first name. For now, option 2 pattern would be sufficient. But I think a more inclusive pattern is needed (like option 3), but I don't understand the use of P141 in this example.

[Christian-McCord]

Nous précisons le genre des Actors (les Constituents dans notre système) en fonction du nom ou du titre de la personne en utilisant un attribut créé à cette fin dans notre thésaurus. Il n'y a que deux possibilités: Female et Male

En ce qui concerne les personnes représentées, nous utilisons des autorités dans la zone Subject. Nous aurions intérêt à référer au Homosaurus Vocabulary car nous ne savons pas toujours quels sont les meilleurs termes. Voici un exemple qui nous a laissé perplexe:

[eecanning]

Hi all, at LINCS we are trying out a pattern similar to the E13 one proposed above for representing information about various identities, including gender. The idea behind this comes out of the CWRC/Orlando ontology, specifically about "Cultural Forms". This pattern we are doing is a translation of that to CRM structure, with some additional changes.

The idea behind this is to highlight how a lot of the terms that we use to describe various forms of identity are loose in definition, and the same term can be used by different people to mean different things. So instead of connecting directly to a concept, you connect to a text label that represents these things. I'll pop in an overview and example diagram (with annotations), but this is the basic idea:
There are differences between
1. somebody making an assertion about someone - such as "Louisa May Alcott is a woman";
2. the relationship that a person holds to that notion - such as Louisa May Alcott's personal relationship to her experience of gender; and
3. the label - "woman" - that is used to describe the identity category.
This allows us to show differing assertions regarding the same identity category that have been made about a person by different biographers, cataloguers, etc., including when the case is one of self-identification or an external identification taking place. It also allows us to show that various identities can be held, including during overlapping periods of time. And lastly, it highlights that the words we use are created and applied in context.

This, like everything else, requires a relationship between structure and vocabularies, and we're working to provide the CWRC vocabularies referenced here as accessible for the shared use. I also highly encourage looking to Homosaurus and other trans metadata projects for vocabularies, especially for the actual label being used (where below in the diagram we have cwrc:woman).

I hope this way of thinking is interesting for you and this discussion!

Diagram 1: Overview

Diagram 2: Example - Lousia May Alcott - Gender

Edit: FYI, the last bit at the bottom is for how the Orlando Project uses annotations (through the Web Annotation Data Model) to connect to source statements, probably less relevant than the rest for this ticket discussion.

[illip]

@eecanning thank you very much for this sharing, I really believe that the proposal that @stephenhart8 and I wanted to make is very close to your proposal.

In particular, we were trying to define this famous E7_Activity, we currently had the idea of going with the fact of "being this gender", but I like the idea that in fact, it is an activity of association with a nomenclature which can, in many cases, be attributed by a third party. I allow myself a few questions, in particular, to nourish the reflection for our next meeting.

1. Do you plan to have a connection between your E7_Activity and the person described? If so, what property are you considering? P11_had_participant, P14_carried_out_by?
2. I'm curious as to why you are using E89_Propositional_Object rather than E55_Type. Is this to allow you to use a term that would not be documented in a thesaurus? If so, how do you plan to indicate the provenance and definition of this E89_Propositional_Object? Do you intend to use another E13_Attribute_Assignment for the "vocabulary" provenance?
3. Does not being able to "withdraw" an attribution seem problematic to you or the fact of managing dates via the E7_Activity seems sufficient?
4. Do you think that a pattern like "belonging to an E74_Group" could be useful? We are not experts but we think that it would be important to document belonging to a group since I can be associated with a gender (and be in agreement with this attribution) without being part of a more or less formal group around the question of this gender.

These are my first questions, but I will continue to think about the matter in the coming weeks, thank you again, and look forward to discussing it face to face soon!

[eecanning]

@illip that's great to hear! I'm exciting that your and @stephenhart8's thinking has been along similar lines. Some quick responses to some excellent questions:

1. We discussed this, and the decision for now is no, but I think it has interesting implications for gender specifically, perhaps more than other "cultural forms" (nationality, religious affiliation, cultural identity, etc.) although this may be because I know more about gender issues than other identity concerns and it presents a way of discussing gender-as-performance, or performing gender. To what extent are we in "active participation" (P14) with a particular label that describes our identity? I think some would argue that we are very active participants in this activity, but that is made more explicit when our selves are at odds with cisheteronormativity - it doesn't mean that the participation in gender is less active when one is conforming, just that the nature of that relationship as being active is disguised by the rest of the structures in our world. As I am one of those people, I would propose this route over P11, but this is definitely requiring discussion.

2. There was a lot of debate about "Type"-ness and what it means to "be typed" when we were working on this that resulted in this. I'm not sure I'm following the second part of your question, I'm sorry.

3. What would be the case for withdrawing an attribution? The idea here is that even though it is known to be wrong, it was still put forward as true by someone, which may have had consequences. the goal would be to provide space to discuss that evolution, or the context from which that incorrect attribution was derived.

4. I do think that a pattern like "belonging to a group" is very interesting, but I'm hesitant about the definition of E74 here. I like it because of the focus on community in that viewing of the question, and something like this could be used to reinforce the notion that identity labels are meaningless outside of community, that an individual cannot "be" without being in relation... this is very much from queer theory, I'm going to cite Samuel Delany for example here, in what it has meant to "come out" - to come into relation, in shared community, with other queer folks. But I don't know if the definition of E74_Group fits with what I am trying to articulate here. Can I ask for clarity on the second part of your question about being (or not being) part of a group around the question of this gender?
   Edit: Coming back to this to say that perhaps this kind of reasoning is why group membership might not in fact be the best route - when we are making statements, often assumptions, about the identities of others, especially people who are deceased, we cannot (should not) place them in community that they did not relate to. It would not be correct to say that somebody was in community during their lifetime due to our assertion of a notion of gender now - particularly if they did not self-identify in the way they are now being described. And now, I will try to stop using this space to think things out!

I do think that part of this is all is that we are using this to discuss association with a wide variety of identity labels, not just gendered ones, and that there are likely nuances for each of those different contexts - there certainly are for gender!
I am also looking forward to getting the chance to talk about all of this :)

[illip]

1. I also prefer the use of P11_had_participant if we want to make the link between the "Person's relationship with label" activity and the person. If we don't think to use P14_carried_out_by would it be better to use E5_Event? I fully understand the problem that this statement can raise, but I have the impression that in certain cases, in particular when it is a question of "self-labeling", that the direct link can become interesting. Obviously, the person can be found through P140 / E13 / P141, but in my opinion, the semantics behind P11_had_participant would be slightly different.

2. I fully understand the issue surrounding typing. Personally, I am also of the opinion that we should limit the use of types or at least be able to manage the origin of these terms and their attributions. I don't know enough about the distinction between E55_Type and E89_Propositional_Object, but I have the impression that the formulation P16/E89 in your example can come very close to P2/E55, I think it would be interesting to highlight the fundamental differences. The second part of my question was about identifying the provenance of the term cwrc: woman. In the event that this is a URI that comes from an external vocabulary, I believe that the provenance will be handled through this vocabulary. However, since E89_Propositional_Object can also be an entity that does not come from a vocabulary, for example, an E33_Linguistic_Object which represents a string, how the provenance would be handled?

3. It is obviously possible to end the activity, but if you want to document the person behind this ending, it's more complicated. If there is a new attribution, there are some logical rules that can be used to assert that the previous one is no longer valid, but I believe that in some cases two attributions can coexist. In addition, what to do in cases where there is no new attribution? There is no solution at the moment within CIDOC CRM, but so much the better if that does not appear to you to be an issue for the moment. :)

4. Again, I believe we have the same considerations. Personally, I would have difficulty with a statement like, “This person joined the “Women” group". I'm not convinced that the E74_Group scope note is ideal for this type of build anyway. But, if we had a way to manage informal groups, it could be a possible avenue if we take care to document the E13_Attribute_Assignment again. However, I remain more in favor of modeling around an individual activity.

Having said that, for formal groups, I believe there is a real benefit to documenting this information in order to be able to distinguish between identity and social involvement. Obviously, like all identity fields, the presence of E13_Attribute_Assignment seems fundamental to me (or at least of the metadata about associated Named Graph) to capture the provenance.

Anyway, look forward to continuing the discussion in a few weeks, thank you very much @eecanning for your contribution. :)

[stephenhart8]

Following the discussions with @eecanning during the Semantic Committee held on 2021/07/15, it has been decided to continue the discussion on this issue. Here are the conclusions (please comment if things have been wrongly interpreted):

The extension CRMsoc, while developing patterns to handle attribution, is still in its early stage of development, and will not be accepted and published for a long time. It has therefore been decided to develop a gender pattern with the tools available without CRMsoc in mind.

It appeared that CHIN’s option n°3, involving an instance of E7_Activity is closer to the pattern developed by LINCS, and should therefore be further investigated:

Option 1 and 2 are too limited and not expressive enough.

Option 4, involving an instance of E74_Group, is interesting as it describes genders as something inherently social, but first the scope note of E74_Group is too restrictive, and second it would be false to always associate an individual to a gender group, as it would always imply belonging to this group.

Even though CHIN’s option 3 and LINCS pattern looks structurally similar, there are some important differences semantically:

• CHIN relies on E55_Type to document concepts, while LINCS has preferred to use E89_Propositional_Object in order to document more fluid concepts (that the meaning can change depending on the culture or time)
• The instance of E7_Activity is the “beingness” of the gender in CHIN’s proposal, while the activity is the relationship between an individual and a gender term in LINCS’ proposal. The concept behind this activity is therefore radically different.
• Because the activity in CHIN’s model is the “beingness” of the gender, a link between the activity and the individual is needed. It could take the form of P14_carried_out_by or P11_had_participant. In LINCS’ perspective, no link is needed between the activity and the individual, as it is not an activity involving the individual. The identification of the person involved is done through the E13_Attribute_Assignment statements.

The main goal of LINCS’ pattern, in addition to the flexibility that both CHIN and LINCS are looking for, is to avoid imposing modern concepts to people of a different time or culture, as well as not presupposing genders, but rather document the association of individuals and genders.

It has been noted during the meeting that the goal of cultural documentation is to record what is publicly stated about someone, and not what could be the private reality of that individual. The goal is therefore to document what has been said about individuals, either false associations or misconceptions. It is then even more important to document the context in which those statements are made.

The next goal for this discussion is to find common grounds within the differences between the two approaches, but also decide to what extent the two approaches (LINCS’ and CHIN’s) should be similar.

[eecanning]

Thank you for inviting me to join for this conversation, and for summarizing here so clearly!
One small clarification - I would say that while at LINCS we do not have a direct P11/P14 between the Actor and the Activity that represents an association with a term that describes a gender identity, it is not because the activity does not involve the individual per se, but that the Activity requires the information that can only come through the E13 - and thus a direct connection from Actor to Activity would undermine the need to provide context that you have articulated. In other words, perhaps - the way that the Actor participates in that Activity is through an act of E13 self-identification; the assertion is the participation (or the participation is the assertion?). A subtle difference, but as we have been discussing the details of the meaning of this pattern, I thought worth mentioning.
It seems to me that this difference in thinking between CHIN and LINCS follows naturally from the different meanings that we brought to that core E7_Activity. I did note some thinking above of why I would propose P14 over P11 if you were to introduce this direct relationship, but for the reasons here LINCS is not planning to go down that path.

Edit: As follow-up to this comment - this article was published today about identity and taxonomy, and contains a passage that I think speaks directly to this (the author applies the same concept to "nonbinary" later on):

When I was eight years old I learned the word bisexual. I wasn’t bisexual until I learned the word bisexual, but I saw the word and read what it meant, and I thought ‘that means the same thing as this inarticulated cluster of feelings and thoughts that I have,’ and that’s how I became bisexual. I felt the same things before and after learning the word, I was the same person on both sides of that definition, but in learning the name of the category I took it on and it became the thing I would call myself.