Re: Replace outdated social models in OWL2 primer

Hi All.
Thank you for all your responses. I'm touched and delighted that the 
community 'cares' and that this isn't a topic that is simply brushed 
aside. I am replying to the original email so as to summarise the 
discussions so far, and identify a path forward.

tldr; let's create a CG, make a document with updated examples, ask W3C 
to adopt it -- preferably with RDF or other relevant WGs, publish a new 
dated version and point the TR/REC link to this.

1) I think there is a general consensus that the examples should be updated.

2) There is some disagreement on the nature of this update and its 
classification in W3C process e.g. is this merely editorial? Here we 
acknowledge that the examples are in a 'REC' document, and that this 
also determines the process for any change/update.

3) Options suggested so far include:

a. directly change the document -- easiest, but not a good option 
because it is a published REC document and it should be maintained

b. publish a new version -- most cumbersome, but probably the best 
option as the history is preserved, we follow w3c best practices re. 
publishing with a different dated version

c. create another separate webpage for examples and link to it -- not 
ideal, the existing examples will still be present and 'canonical' and 
we'll be saying this is REC but don't use it which is not good practice

4) From the above, I think option (b) is the most logical choice. The 
discussion after this reflected the process by which this new version 
can be published -- whether we can classify this as a minor change (less 
redtape, quicker) or this classifies as a significant change (has to go 
through CG/WG and approval process).

5) So from the above, I suggest we create the CG, then within it create 
a mirror document of the OWL primer with updated examples, and suggest 
it back to the W3C for incorporation. It will need to go through a WG 
(AFAIK) -- and I hope one of the existing ones picks this up to push it 
ahead (e.g. RDF-star WG would be closest?).

6) The CG is also a good option, because we can also see if there are 
other documents with similar issues - and if so, what are they and what 
should be the fix. At the same time, we can recommend some inclusivity 
principles / codes of decency to be added to existing W3C guidelines for 
how to create examples. We don't want everyone to read them before 
creating every single example, but they should be helpful if such an 
issue arises again elsewhere.

7) The CG is also again a good option as the W3C is currently exploring 
how to promote CGs outputs and how to smooth the process of uptaking CG 
work within W3C (WG, etc.). So this will also be helpful for other CGs 
and communities by providing a path to test the work done here so far.

8) Why CG and not a WG: With the CG, anyone can participate, we get a 
github repo to track this work (formally), and its all in w3c. A WG is 
also a good option, but I am not sure whether updating a single document 
with examples should be a new WG.

Regards, Harsh

On 15/03/2025 18:32, Harshvardhan J. Pandit wrote:
> Hi All.
> While revisiting the OWL2 primer recently at https://www.w3.org/TR/owl2- 
> primer/, I found several examples for showing how OWL2 works that try to 
> model social constructs like man/woman, parent/child, father/mother in a 
> way that I consider increasingly out of touch with today. I propose that 
> these be changed to something that has no issues or over which no 
> social, ethical, or political discussions are necessary for the adopter 
> as the goal here is to show how OWL2 works.
> 
> ---
> 
> E.g. Sec 4.2 Suppose we also want to state that all mothers are women: 
> SubClassOf( :Mother :Woman )
> 
> Here, it represents that mother is a strict subset of woman i.e. only 
> women can be mothers. However, "Woman" here is referring to "woman as a 
> human of female sex" and not "woman as gender". Rather than get into 
> what these definitions should be, or what kind of sets exist and their 
> intersections (e.g. woman, trans-woman, trans-man, intersex, and so on) 
> - my point is that these are not good examples to start modelling with 
> even if they might have been seen as "intuitive" some decades ago.
> 
> ---
> 
> E.g. Sec 4.3 For example, if we consider the classes Man and Woman, we 
> know that no individual can be an instance of both classes (for the sake 
> of the example, we disregard biological borderline cases)... 
> DisjointClasses( :Woman :Man )
> 
> Again, we should not exclude anyone here just because they are 'on the 
> fringes' and also because there are ways people can change their sex and 
> their gender -- so this example is not a good example to use here.
> 
> ---
> 
> E.g. Sec 4.6 For instance, the statement that B is the wife of A 
> obviously implies that B is a woman while A is a man. 
> ObjectPropertyDomain( :hasWife :Man ) 
> ObjectPropertyRange( :hasWife :Woman ) ... Having these two axioms in 
> place and given e.g. the information that Sasha is related to Hillary 
> via the property hasWife, a reasoner would be able to infer that Sasha 
> is a man and Hillary a woman.
> 
> While I don't know what is the canonical name for people who are not 
> married (partner?) or who are in a same-sex/gender relationship -- this 
> is again a good point to note that the example has implications beyond 
> OWL and shouldn't be used here.
> 
> ---
> 
> E.g. Sec 5.1 The following example states that the class Mother consists 
> of exactly those objects which are instances of both Woman and Parent 
> EquivalentClasses(
>     :Mother
>     ObjectIntersectionOf( :Woman :Parent )
>   )
> 
> Again, this has more implications to consider such as transgender 
> mothers and also motherhood following sex-change. Therefore, this is not 
> a good example to learn about how OWL.
> 
> We also have in Sec 10
> SubClassOf(
>     :Father
>     ObjectIntersectionOf( :Man :Parent )
>   )
> 
> ---
> 
> E.g. Sec 5.1 we could characterize the class of all parents as the union 
> of the classes Mother and Father
> EquivalentClasses(
>     :Parent
>     ObjectUnionOf( :Mother :Father )
>   )
> 
> Parents are not exclusive to mothers and fathers e.g. stepmother or 
> grandparent, or even non-biological parents (though they would be called 
> the same). Further, it might be seen as saying parents are always a 
> combination of a mother and a father - though this is not in the text or 
> the rule. (I'll note that in Sec.9 the concept "SocialRole" is stated as 
> a metaclass of Father, but isn't defined or explained)
> 
> ---
> 
> Is this change urgent? No. Is this outright offending anyone? I don't 
> think so. But should we change this? Yes, I think so. Each year there 
> will be many more new people and newer generations learning OWL, and 
> many of us relearning it. So we shouldn't wait for this to be an issue 
> either for being out of touch or for not being considerate before we 
> change it.
> 
> So what do we change this with? I think examples with animals (cats, 
> dogs), shapes, etc. are universal, and aren't at risk of not conforming 
> to society or for not being empathic. Or if we still want to model 
> people, let's do friendships and work relationships that have no 
> personal characteristics. For OWL specifically, I think the Pizza 
> ontology used as a tutorial in Protege is also a good option as 
> everybody likes pizza! (well, I hope).

-- 
---
Harshvardhan J. Pandit, Ph.D
Assistant Professor
ADAPT Centre, Dublin City University
https://harshp.com/

Received on Monday, 24 March 2025 10:49:15 UTC