Kingsley Idehen | 5 Feb 2010 20:31
Picon

Re: [foaf-dev] works relation

Story Henry wrote:
> On 5 Feb 2010, at 20:09, Kingsley Idehen wrote:
>
>   
>> Story Henry wrote:
>>     
>>> There are people in foaf, and companies. But there is no way to say that a person works in a company it
seems. What should one do there?
>>>
>>> Henry
>>>
>>> Social Web Architect
>>> http://bblfish.net/
>>>
>>> _______________________________________________
>>> foaf-dev mailing list
>>> foaf-dev@...
>>> http://lists.foaf-project.org/mailman/listinfo/foaf-dev
>>>
>>>  
>>>       
>> All,
>>
>> See: http://vocab.org/relationship/.html
>>
>> Its linked to FOAF.
>>
>> This ontology saved me a ton of time when trying to verify some reasoner enhancements to Virtuoso re:
>> owl:SymmetricalProperty, owl:TransitiveProperty, and owl:inverseOf.
>>
(Continue reading)

Story Henry | 5 Feb 2010 20:38

Re: [foaf-dev] works relation


On 5 Feb 2010, at 20:31, Kingsley Idehen wrote:

>> 
>> yes, except that this ontology is pretty crap. 
>> I have pointed out a few years ago that it does not make sense, and I have seen no action to improve it. Take
just for exampel:
>> 
>> <wouldLikeToKnow>     a rdf:Property;
>>         :domain <http://xmlns.com/foaf/0.1/Person>;
>>         :isDefinedBy <>;
>>         :label "would like to now" <at> en;
>>         :range <http://xmlns.com/foaf/0.1/Person>;
>>         :subPropertyOf <http://xmlns.com/foaf/0.1/knows>;
>>         skos:definition "A person whom this person would desire to know more closely." <at> en .
>> 
>> from this if you wouldLikeToKnow someone, then you know them.
>> Which is nonsensical. So I wouldLikeToKnow SamanthaFox, therefore I know her! 
>>  
> Okay, scrap that relation.
> 
> I focused on the employment terms, albeit for my own selfish purposes i.e., verifying new reasoner
features in Virtuoso  :-)

Don't get me wrong. I like the names of all those relations, but until they are a bit surer they should remove
any of the relations that have entailment consequences. But they seem completely unresponsive to
feedback, as far as I can see... So they should be avoided.

Henry

(Continue reading)

Kingsley Idehen | 5 Feb 2010 20:41
Picon

Re: [foaf-dev] works relation

Story Henry wrote:
> On 5 Feb 2010, at 20:31, Kingsley Idehen wrote:
>
>   
>>> yes, except that this ontology is pretty crap. 
>>> I have pointed out a few years ago that it does not make sense, and I have seen no action to improve it. Take
just for exampel:
>>>
>>> <wouldLikeToKnow>     a rdf:Property;
>>>         :domain <http://xmlns.com/foaf/0.1/Person>;
>>>         :isDefinedBy <>;
>>>         :label "would like to now" <at> en;
>>>         :range <http://xmlns.com/foaf/0.1/Person>;
>>>         :subPropertyOf <http://xmlns.com/foaf/0.1/knows>;
>>>         skos:definition "A person whom this person would desire to know more closely." <at> en .
>>>
>>> from this if you wouldLikeToKnow someone, then you know them.
>>> Which is nonsensical. So I wouldLikeToKnow SamanthaFox, therefore I know her! 
>>>  
>>>       
>> Okay, scrap that relation.
>>
>> I focused on the employment terms, albeit for my own selfish purposes i.e., verifying new reasoner
features in Virtuoso  :-)
>>     
>
> Don't get me wrong. I like the names of all those relations, but until they are a bit surer they should remove
any of the relations that have entailment consequences. But they seem completely unresponsive to
feedback, as far as I can see... So they should be avoided.
>   
(Continue reading)

Story Henry | 5 Feb 2010 20:43

Re: [foaf-dev] works relation

On 5 Feb 2010, at 20:41, Kingsley Idehen wrote:

> Story Henry wrote:
>> On 5 Feb 2010, at 20:31, Kingsley Idehen wrote:
>> 
>>  
>>>> yes, except that this ontology is pretty crap. I have pointed out a few years ago that it does not make
sense, and I have seen no action to improve it. Take just for exampel:
>>>> 
>>>> <wouldLikeToKnow>     a rdf:Property;
>>>>        :domain <http://xmlns.com/foaf/0.1/Person>;
>>>>        :isDefinedBy <>;
>>>>        :label "would like to now" <at> en;
>>>>        :range <http://xmlns.com/foaf/0.1/Person>;
>>>>        :subPropertyOf <http://xmlns.com/foaf/0.1/knows>;
>>>>        skos:definition "A person whom this person would desire to know more closely." <at> en .
>>>> 
>>>> from this if you wouldLikeToKnow someone, then you know them.
>>>> Which is nonsensical. So I wouldLikeToKnow SamanthaFox, therefore I know her!        
>>> Okay, scrap that relation.
>>> 
>>> I focused on the employment terms, albeit for my own selfish purposes i.e., verifying new reasoner
features in Virtuoso  :-)
>>>    
>> 
>> Don't get me wrong. I like the names of all those relations, but until they are a bit surer they should
remove any of the relations that have entailment consequences. But they seem completely unresponsive to
feedback, as far as I can see... So they should be avoided.
>>  
> Sure, that why "reasoning act" should be subjective re. SPARQL.
(Continue reading)

Kingsley Idehen | 5 Feb 2010 20:47
Picon

Re: [foaf-dev] works relation

Story Henry wrote:
> On 5 Feb 2010, at 20:41, Kingsley Idehen wrote:
>
>   
>> Story Henry wrote:
>>     
>>> On 5 Feb 2010, at 20:31, Kingsley Idehen wrote:
>>>
>>>  
>>>       
>>>>> yes, except that this ontology is pretty crap. I have pointed out a few years ago that it does not make
sense, and I have seen no action to improve it. Take just for exampel:
>>>>>
>>>>> <wouldLikeToKnow>     a rdf:Property;
>>>>>        :domain <http://xmlns.com/foaf/0.1/Person>;
>>>>>        :isDefinedBy <>;
>>>>>        :label "would like to now" <at> en;
>>>>>        :range <http://xmlns.com/foaf/0.1/Person>;
>>>>>        :subPropertyOf <http://xmlns.com/foaf/0.1/knows>;
>>>>>        skos:definition "A person whom this person would desire to know more closely." <at> en .
>>>>>
>>>>> from this if you wouldLikeToKnow someone, then you know them.
>>>>> Which is nonsensical. So I wouldLikeToKnow SamanthaFox, therefore I know her!        
>>>>>           
>>>> Okay, scrap that relation.
>>>>
>>>> I focused on the employment terms, albeit for my own selfish purposes i.e., verifying new reasoner
features in Virtuoso  :-)
>>>>    
>>>>         
(Continue reading)

Story Henry | 5 Feb 2010 20:51

Re: [foaf-dev] works relation


On 5 Feb 2010, at 20:47, Kingsley Idehen wrote:

> Story Henry wrote:
>> On 5 Feb 2010, at 20:41, Kingsley Idehen wrote:
>> 
>>  
>>> Story Henry wrote:
>>>    
>>>> On 5 Feb 2010, at 20:31, Kingsley Idehen wrote:
>>>> 
>>>>       
>>>>>> yes, except that this ontology is pretty crap. I have pointed out a few years ago that it does not make
sense, and I have seen no action to improve it. Take just for exampel:
>>>>>> 
>>>>>> <wouldLikeToKnow>     a rdf:Property;
>>>>>>       :domain <http://xmlns.com/foaf/0.1/Person>;
>>>>>>       :isDefinedBy <>;
>>>>>>       :label "would like to now" <at> en;
>>>>>>       :range <http://xmlns.com/foaf/0.1/Person>;
>>>>>>       :subPropertyOf <http://xmlns.com/foaf/0.1/knows>;
>>>>>>       skos:definition "A person whom this person would desire to know more closely." <at> en .
>>>>>> 
>>>>>> from this if you wouldLikeToKnow someone, then you know them.
>>>>>> Which is nonsensical. So I wouldLikeToKnow SamanthaFox, therefore I know her!                  
>>>>> Okay, scrap that relation.
>>>>> 
>>>>> I focused on the employment terms, albeit for my own selfish purposes i.e., verifying new reasoner
features in Virtuoso  :-)
>>>>>           
(Continue reading)

Danny Ayers | 5 Feb 2010 21:43
Picon
Gravatar

Re: [foaf-dev] works relation

On 5 February 2010 20:51, Story Henry <henry.story@...> wrote:

> In vi I can edit that too pretty easily. The point is that if the people publishing an ontology can't be
relied upon to take the community feedback seriously, then one should consider that ontology dead.
Otherwise how would one decide in the future which ontology to use? People could delte one relation they
don't consider important, others other rels. And then I can just see how easy it is going to be to go to
companies and argue about what is correct at a later date.
> Deleting relations from ontologies, should be a case of force majeur. Serious stuff. Not reasoning on
certain things is ok, for speed reasons, but not because one disagrees with the ontology.

The definition of the wouldLikeToKnow property certainly does seem
faulty, but as a pragmatic means to an end I can't see a problem with
using the term under relevant circumstances, or even just being
selective about which terms to use from a given ontology and what
kinds of reasoning to apply over their definitions. Not taking on
board community feedback is no doubt a heinous crime, but one I'm sure
we're all guilty of... But then again, if you're not interested in
RDFS inference, there isn't really that much amiss with the
definitions. (Plenty of vocabularies are OWL Full rather than DL after
all, where do we set our sights..?).

To appeal to your philosophical side, isn't this just a generalisation
of the idea of ontological commitment? Surely we can pick & choose
which statements out there are useful for our own purposes? (Whether
they come from datasets or ontologies).

It is untidy though, but I can't think of a tidy solution - perhaps as
another party, treat the ontology as a named graph (an owl:Ontology no
less), then make third-party annotations on it? (like you might do on
a mailing list :)
(Continue reading)

Story Henry | 5 Feb 2010 21:49

Re: [foaf-dev] works relation


On 5 Feb 2010, at 21:43, Danny Ayers wrote:
> It is untidy though, but I can't think of a tidy solution - perhaps as
> another party, treat the ontology as a named graph (an owl:Ontology no
> less), then make third-party annotations on it? (like you might do on
> a mailing list :)
> 
> Dunno, it just seems broken in principle to take a black/white
> approach to vocabularies out there - most after all are evolving
> anyway, bad terms being just a special case of messy data.

Well all that will happen is that I will say it's a bad ontology to anyone who asks me, and that people should
not use it - until some of the feedback given a couple of years ago is taken into account. Or perhaps I could be
shown to be wrong. 

But that is a one page ontology. If we can't be critical at that level anymore then we are just setting
ourselves up for being dismissed in most places. 

We need many of these relationships. Perhaps Ian could just fix some of these and I'd be happy to support it.

Henry
Story Henry | 5 Feb 2010 21:56

Re: [foaf-dev] works relation

Well it was a bit less than 2 years ago that we discussed this on this list:

http://lists.foaf-project.org/pipermail/foaf-dev/2008-July/009265.html

Henry

On 5 Feb 2010, at 21:49, Story Henry wrote:

> 
> On 5 Feb 2010, at 21:43, Danny Ayers wrote:
>> It is untidy though, but I can't think of a tidy solution - perhaps as
>> another party, treat the ontology as a named graph (an owl:Ontology no
>> less), then make third-party annotations on it? (like you might do on
>> a mailing list :)
>> 
>> Dunno, it just seems broken in principle to take a black/white
>> approach to vocabularies out there - most after all are evolving
>> anyway, bad terms being just a special case of messy data.
> 
> 
> Well all that will happen is that I will say it's a bad ontology to anyone who asks me, and that people should
not use it - until some of the feedback given a couple of years ago is taken into account. Or perhaps I could be
shown to be wrong. 
> 
> But that is a one page ontology. If we can't be critical at that level anymore then we are just setting
ourselves up for being dismissed in most places. 
> 
> We need many of these relationships. Perhaps Ian could just fix some of these and I'd be happy to support it.
> 
> Henry
(Continue reading)

Danny Ayers | 5 Feb 2010 21:59
Picon
Gravatar

Re: [foaf-dev] works relation

Turning it around, if the terms are likely to be useful, it's worth
grumbling again 2 years on :)

On 5 February 2010 21:56, Story Henry <henry.story@...> wrote:
> Well it was a bit less than 2 years ago that we discussed this on this list:
>
> http://lists.foaf-project.org/pipermail/foaf-dev/2008-July/009265.html
>
> Henry
>
> On 5 Feb 2010, at 21:49, Story Henry wrote:
>
>>
>> On 5 Feb 2010, at 21:43, Danny Ayers wrote:
>>> It is untidy though, but I can't think of a tidy solution - perhaps as
>>> another party, treat the ontology as a named graph (an owl:Ontology no
>>> less), then make third-party annotations on it? (like you might do on
>>> a mailing list :)
>>>
>>> Dunno, it just seems broken in principle to take a black/white
>>> approach to vocabularies out there - most after all are evolving
>>> anyway, bad terms being just a special case of messy data.
>>
>>
>> Well all that will happen is that I will say it's a bad ontology to anyone who asks me, and that people should
not use it - until some of the feedback given a couple of years ago is taken into account. Or perhaps I could be
shown to be wrong.
>>
>> But that is a one page ontology. If we can't be critical at that level anymore then we are just setting
ourselves up for being dismissed in most places.
(Continue reading)

Story Henry | 5 Feb 2010 22:04

Re: [foaf-dev] works relation


On 5 Feb 2010, at 21:59, Danny Ayers wrote:

> Turning it around, if the terms are likely to be useful, it's worth
> grumbling again 2 years on :)

The problem is a lot of people do copy and paste. So if we support code that works that way, then we also may
spread a bad habit. It's a bit like a disease. So please just fix it now! I'd say just remove the problematic
relationships. Comment them out. Let's come back to it later when we see how people really use the vocab.

Henry

> 
> On 5 February 2010 21:56, Story Henry <henry.story@...> wrote:
>> Well it was a bit less than 2 years ago that we discussed this on this list:
>> 
>> http://lists.foaf-project.org/pipermail/foaf-dev/2008-July/009265.html
> 
Peter Williams | 6 Feb 2010 01:14
Favicon

Re: [foaf-dev] works relation

>>>>> from this if you wouldLikeToKnow someone, then you know them.
>>>>> Which is nonsensical. So I wouldLikeToKnow SamanthaFox, therefore I know her!        

Nah. It perfectly sensible. People are more subtle than any formal logic. (For most of the world's male
foaf-dev readers, Sam Fox was a titty bar sensation, who worked the media to make herself rich and famous).

I remember meeting the brothers of my first girlfriend (the cutest brownskinned 18year old English Moslem
girl on the entire planet, ex Burma). As schoolboys, they looked at me (age about 15 and 17) asking: so do you
know our sister (in the biblical sense of "knows")? The answer was actually no, but given their sister's
overt lustiness, that would have been hard to believe of any normal male. 

As brothers from a good family trying properly to decide it they'd want me long term in their family circle
(assuming I'd demonstrated respect for their sister and therefore any other female person), they did
what any slightly conservative family member would do, when qualifying "knows". It came totally
naturally to me then and nowto apply such _parochial_ qualifiers to "knows". It was a matter of trust formation.

Speaking as a programmer it makes total sense to reuse knows, when one sets up a "provisional"  "would like
to" know. I definitely wanted to know that girl (we were ages 17 and 18 respectively). At the outset I could
not have put here on the actual knows list (even 20 years ago, before the era of social networks). An
immature male, Id have wanted to send 2 signals: (a) I want to get the point where we are both comfy using that
label (the "so you publicly have a  boyfriend, and I'm it!" moment), and (b) meantime, I'd want to qualify
that we are not there yet.

The point about trust is that is always transcends the public/private line that delineates the worlds of
public and private. Trust is not the same as the rigid compartmentalization of authorization schemes.
It's the rules that allow the breaking of the rules (RTHATBR), as the private and public space interact to
meet both the needs of society AND (a couple of) individuals.

-----Original Message-----
From: foaf-dev-bounces@...
(Continue reading)

Story Henry | 6 Feb 2010 11:32

Re: [foaf-dev] works relation


On 6 Feb 2010, at 01:14, Peter Williams wrote:

>>>>>> from this if you wouldLikeToKnow someone, then you know them.
>>>>>> Which is nonsensical. So I wouldLikeToKnow SamanthaFox, therefore I know her!        
> 
> Nah. It perfectly sensible. People are more subtle than any formal logic. (For most of the world's male
foaf-dev readers, Sam Fox was a titty bar sensation, who worked the media to make herself rich and famous).
> 
> I remember meeting the brothers of my first girlfriend (the cutest brownskinned 18year old English
Moslem girl on the entire planet, ex Burma). As schoolboys, they looked at me (age about 15 and 17) asking:
so do you know our sister (in the biblical sense of "knows")? The answer was actually no, but given their
sister's overt lustiness, that would have been hard to believe of any normal male. 
> 
> As brothers from a good family trying properly to decide it they'd want me long term in their family circle
(assuming I'd demonstrated respect for their sister and therefore any other female person), they did
what any slightly conservative family member would do, when qualifying "knows". It came totally
naturally to me then and nowto apply such _parochial_ qualifiers to "knows". It was a matter of trust formation.
> 

That's a nice story. 

But foaf:knows refers to a very generic knows relation. By you (:pw) meeting her (:g) in person, as you
describe it above - calling her your girlfriend - it was true that:

 <at> prefix foaf: <http://xmlns.com/foaf/0.1/> .

:pw foaf:knows :g .

When they use the word "knows", you understood that to be a label for a very different relation
(Continue reading)

Story Henry | 6 Feb 2010 11:57

Re: [foaf-dev] works relation


On 6 Feb 2010, at 11:32, Story Henry wrote:

> In any case to come back to  rel:wouldLikeToKnow. If the documentation really wants to have it have the
meaning of being a subrelation of foaf:knows, then it should state so in the human readable
documentation. Currently the rdfs:comment is the one people are bound to read and it will make people use
it in a way different from the way the ontology specifies. Perhaps it should say not "would like to know" but
"would like to know better". Then that would be ok. But currently there is a gap between what the machine
definition says and what the human spec says. And this is bound to lead to misunderstandings all over the
place. 

And even then it would not work.

foaf:knows is specified as being a symmetrical relation. This is not done formally, but in the text. As a
result, rel:wouldLikeToKnow should also be a symmetrical relation, even if only in the informal way
described by foaf.  But clearly "would like to know better" is NOT symmetrical. It is quite clear that I can
express the desire to know someone better without them having the same desire in return.

Henry
Story Henry | 6 Feb 2010 12:00

Re: [foaf-dev] works relation


On 6 Feb 2010, at 11:57, Story Henry wrote:

> 
> On 6 Feb 2010, at 11:32, Story Henry wrote:
> 
>> In any case to come back to  rel:wouldLikeToKnow. If the documentation really wants to have it have the
meaning of being a subrelation of foaf:knows, then it should state so in the human readable
documentation. Currently the rdfs:comment is the one people are bound to read and it will make people use
it in a way different from the way the ontology specifies. Perhaps it should say not "would like to know" but
"would like to know better". Then that would be ok. But currently there is a gap between what the machine
definition says and what the human spec says. And this is bound to lead to misunderstandings all over the
place. 
> 
> And even then it would not work.
> 
> foaf:knows is specified as being a symmetrical relation. This is not done formally, but in the text. As a
result, rel:wouldLikeToKnow should also be a symmetrical relation, even if only in the informal way
described by foaf.  But clearly "would like to know better" is NOT symmetrical. It is quite clear that I can
express the desire to know someone better without them having the same desire in return.

Mhh, I'll take that back. All I think that is required is that foaf:knows be symmetrical.
It is not required that subrelations also be symmetrical. (I think I made that mistake before)

Sorry.

Henry

> 
> Henry
(Continue reading)

Story Henry | 6 Feb 2010 14:21

Re: [foaf-dev] works relation

On 6 Feb 2010, at 11:32, Story Henry wrote:

> 
> In any case to come back to  rel:wouldLikeToKnow. If the documentation really wants to have it have the
meaning of being a subrelation of foaf:knows, then it should state so in the human readable
documentation. Currently the rdfs:comment is the one people are bound to read and it will make people use
it in a way different from the way the ontology specifies. Perhaps it should say not "would like to know" but
"would like to know better". Then that would be ok. 

I need to be a bit clearer here. The definition of wouldLikeToKnow is

    <wouldLikeToKnow>     a rdf:Property;
         :domain <http://xmlns.com/foaf/0.1/Person>;
         :isDefinedBy <>;
         :label "would like to now" <at> en;
         :range <http://xmlns.com/foaf/0.1/Person>;
         :subPropertyOf <http://xmlns.com/foaf/0.1/knows>;
         skos:definition "A person whom this person would desire to know more closely." <at> en .

So it does say "would desire to know _more_closely_" . 

But my guess is that what people really want to use that wouldLikeToKnow relationship for is for getting to
know people in such a way that the other knows them too. Ie: this is request to be put in contact with someone,
of whom one has heard, and whom one knows of (but not foaf:knows). 

The problem is that the relationship ontology seems to think that foaf:knows is the most generic
relationship between two people. It is not. The definition of foaf:knows specifies quite clearly that
the relationship is symmetrical:

"
(Continue reading)


Gmane