Re: Open Issues in the Model

Dear all,

I tent to agree with Francesca in that I see antonymy as something 
different from variation as originally conceived and understood in our 
papers. I think our starting point was the comparison of variants at 
three levels: form, sense, ontology concept; and the underlying idea was 
that variants do not incur in a difference in meaning (same ontology 
entity) or have very closely related meanings (if we consider translation).
As Francesca says, antonomy is a semantic relation, somewhere in between 
the lexicon and the ontology.

However, in an attempt to accomodate John's example, since I understand 
that someone may want to say that /do /and /undo /are related, I have 
come up with a different classification.

In the attached document, I include two slides with two classification 
trees. The 1st one represents the way in which we originally understood 
variation.
The 2nd one proposes a new classification. Slides 3 and 4 cointain some 
drawings that helped us here at UPM to understand variation in the  
context of ontolex.

I do not know if you share my view. Maybe I am completely wrong or 
missing some important points, or maybe our missunderstandigs are due to 
our different backgrounds.
Hopefully we come to an agreement soon... :)

I hope to be able to attend the telco today, but I have a medical 
appointment at 14... :(
Elena

El 14/10/2014 16:07, Francesca Frontini escribió:
> Hi everyone,
> sorry to enter so abruptly in the discussion. I was trying to follow 
> the  last confcall but my line was very bad (I was calling via Skype)...
>
> From a lexicographic perspective I really feel uncomfortable about 
> this idea of defining things such as  "antonymy" under the umbrella of 
> "variants".
> Variants are generally just FormVariants in linguistics, as far as i 
> remember: relations like antonymy and synonymy are just good old 
> *semantic* relations for me.
> I understand that in Ontolex the *semantics* should be on the ontology 
> side, but then this may cause ANY relation among senses on the lexicon 
> side to be a bit.... out of place, maybe?
>
> As for other phenomena, what about derivation, ... would that fall 
> under LexicalVariant?
>
> I'm sure you have a good reason for that, and if I read all the 
> documentation I would find it, but I thought it is quicker to just ask :)
>
> Regards,
> Francesca
>
> 2014-10-14 15:22 GMT+02:00 John P. McCrae 
> <jmccrae@cit-ec.uni-bielefeld.de 
> <mailto:jmccrae@cit-ec.uni-bielefeld.de>>:
>
>
>
>     On Tue, Oct 14, 2014 at 3:06 PM, Jorge Gracia <jgracia@fi.upm.es
>     <mailto:jgracia@fi.upm.es>> wrote:
>
>         Hi John/all,
>
>         Here you are my comments on some of your reported issues...
>
>         2014-10-10 20:07 GMT+02:00 John P. McCrae
>         <jmccrae@cit-ec.uni-bielefeld.de
>         <mailto:jmccrae@cit-ec.uni-bielefeld.de>>:
>
>             Core:
>
>               * We could/should consider using dct:language instead of
>                 ontolex:languageURI
>
>         Yes, I would be in favour. The drawback, though, is the
>         redundancy of names we would have: ontolex:language (for
>         string languages) and dct:language (for URI languages). I
>         think that is the reason why we introduced "languageURI"
>
>             Variation
>
>               * Lexical Variant is defined between either forms /or/
>                 lexical entries... there should be a class that is
>                 only for forms and a class that is only for entries
>
>         What about TerminologicalVariant (for senses), LexicalVariant
>         (for entries), and FormVariant (for forms) ?
>         Or even simpler!: SenseVariant, EntityVariant, FormVariant
>
>     Yes!!!!
>
>               * All variants are specified only in their 'reified'
>                 form, do we want to allow users to directly state
>                 variation between two entries (or forms or senses)
>                 with a single triple?
>
>         A possible option is to use OWL2 "punning"
>         (http://www.w3.org/TR/owl2-new-features/#F12:_Punning),
>         although I am not familiarised with it and I do not control
>         their possible implications well
>
>     Yeah the issue is that we probably don't want to pun the classes
>     as properties.. we want to be able to say something like this
>
>     :sense1 lexinfo:antonym :sense2 .
>
>     Where
>
>     lexinfo:antonym rdfs:subPropertyOf vartrans:senseVariant
>
>     Currently we have to do the following:
>
>     :antonym1 a vartrans:Variant ;
>       vartrans:source :sense1 ;
>       vartrans:target :sense2 ;
>       vartrans:category lexinfo:antonym .
>
>     Regards,
>     John
>
>               * Are the Interlingual-/IntralingualVariant classes
>                 necessary?
>
>         I think we already decided in the last telco to remove them,
>         as all the possible variants are already covered by the other
>         types. Am I right?
>
>             Metadata
>
>               * The Lexicon class is a duplicate of one already in the
>                 core
>
>         In any case I would keep it in the core as a first class
>         citizen (and I see no reason why reusing it in other modules,
>         such as the "metadata" one, would not be possible)
>
>         Regards,
>
>         Jorge
>
>
>
>         -- 
>         Jorge Gracia, PhD
>         Ontology Engineering Group
>         Artificial Intelligence Department
>         Universidad Politécnica de Madrid
>         http://jogracia.url.ph/web/
>
>
>