[Top] [All Lists]

Re: [ontolog-forum] Consensus on labeling of relationships?

To: ontolog-forum@xxxxxxxxxxxxxxxx
From: John F Sowa <sowa@xxxxxxxxxxx>
Date: Tue, 21 May 2013 12:23:35 -0400
Message-id: <519B9F87.6090603@xxxxxxxxxxx>
Phil and Tom,    (01)

> Is there any consensus in the KR community on best practices
> for labeling relationships among objects?
> And a closely related question: Why English?    (02)

Short answers:    (03)

   For #1, no, but there are frequently used options.    (04)

   For #2, because the people who are writing most of the documents
   about the topic are writing in English.    (05)

> The sentence _Christine has [a disease]._ is a kind of circumlocution or
> shorthand, and it is a very typical construction in English. But Christine
> does not "have" a disease in the same sense that she "has" a
> pencil. That facile ambiguity makes me uncomfortable with using English
> _to have_ in the human-readable labels for expressions of relationships.    (06)

That is not a problem.  It's a solution that can be adapted to many
different languages.    (07)

In English and many other languages, the preposition 'of' (or its
equivalent in other languages), the possessive or genitive case
in many languages, and the verb 'have' (or its equivalents in many
languages) are related to one another by simple transformations.    (08)

Depending on your point of view, 'of' and 'have' can be considered
highly ambiguous, or they can be considered to have one simple
core meaning: they show that there exists some relation between
two noun phrases.  For noun phrases N1 and N2, the following
transformations are usually possible (and it's also possible
to classify and explain the exceptions):    (09)

    "N1 of N2"  <=>  "N2 has N1"  <=>  "N2's N1"    (010)

Therefore, a widely used convention is to append 'has' to
a noun that names the relation in question:  hasDisease(x,y)
or hasPossession(x,y).    (011)

The inverse relation, if desired, can be named by appending
'is' in front and 'of' at the end:  isDiseaseOf(y,x) or
isPossessionOf(y,x).    (012)

Similar conventions and transformations can be adapted
to other languages.    (013)

> Concepts themselves
>    a.) have language relations that allow them to be identified in
>        multiple languages, coding systems, even pictures or sound,
>    b.) are itself semantic net-lets.
> We can formulate a relation as phrases in the languages selected
> by the user.    (014)

I agree.  That method has been widely used for the past 50+ years
in artificial intelligence and computational linguistics.  I don't
know why the NeuroCollective is being secretive.  But if they're
planning to patent the idea, there is a huge amount of prior art.    (015)

John    (016)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J    (017)

<Prev in Thread] Current Thread [Next in Thread>