W3C home > Mailing lists > Public > public-ontolex@w3.org > April 2017

Re: about the definition of LexicalConcept in the specs

From: John McCrae <john@mccr.ae>
Date: Thu, 20 Apr 2017 11:56:16 +0100
Message-ID: <CAC5njqpNhzLJJWqMAii9gQnyJ5y-Anx1ZqaDx1Nwu+_5WkJvVQ@mail.gmail.com>
To: Armando Stellato <stellato@uniroma2.it>
Cc: "public-ontolex@w3.org" <public-ontolex@w3.org>
Hi Armando,

An interesting point. Yes, the definition is quite inexact and the
intention of "can be lexicalized" here was simply to state that such words
can be used to express the word in natural language and vica versa. I don't
think we intended to consider the process by which the concept is created.

Moreover, we should keep the definition quite loose, as we wish lexical
concept to fit the idea of synset, as you said. As such, it should be noted
that for many wordnets created for languages other than English, there are
gaps where the English synset is not lexicalized in the target language but
they still exist in the hierarchy, hence they are unlexicalized lexical
concepts, so the semasialogical/onomasiological distinction does not quite
fit in all cases.

The distinction between concepts and references is really meant to be a
question of the purpose of the resource: concepts are created as part of
the lexicon, where as references are created to describe a domain. In fact,
in the case of onomasiological lexicons (where the concepts drive the
lexicography, an example is the SALDO lexicon), then the concepts are still
lexical concepts rather than references.

Regards,
John

On Wed, Apr 19, 2017 at 10:29 AM, Armando Stellato <stellato@uniroma2.it>
wrote:

> Dear all,
>
>
>
> I was caught by the deskmate syndrome: while explaining something about
> Ontolex to a colleague, had the occasion to revise the specs and noticed
> something that doesn’t add up to me.
>
>
>
> In particular, I think the definition of LexicalConcept goes against its
> original intention (and intension).
>
>
>
> Taken from the specs in: http://www.w3.org/2016/05/
> ontolex/#lexical-concept
>
>
>
> “[…] sometimes we would like to express the fact that a certain lexical
> entry evokes <http://www.w3.org/2016/05/ontolex/#evokes> a certain mental
> concept rather than that it refers to a class with a formal interpretation
> in some model. Thus, in lemon we introduce the class Lexical Concept
> <http://www.w3.org/2016/05/ontolex/#LexicalConcept> that represents *a
> mental abstraction, concept or unit of thought that can be lexicalized by a
> given collection of senses*.
>
>
>
> I disagree on the “can be lexicalized”.
>
>
>
> The LexicalConcept was introduced to express the very nature of those
> elements that are created with a semasiological approach, that is: you have
> a lexical expression, and you create a concept representing the meaning of
> that word. Concepts created with an onomasiological approach (i.e. I know
> XXX exists, and then I look for the words referring to it), as in thesauri,
> are not LexicalConcepts. I remember that I suggested the name
> LexicalConcept, borrowing it from Miller [1], with a clear hint to WordNet
> synsets (but providing a more general name), which are created a-posteriori
> to provide meanings to the words in the WordNet lexical database.
>
>
>
> Thus, a synset in WordNet is a ontolex:LexicalConcept, a concept in
> Agrovoc is (should be?) just a skos:Concept, and the we should invert the
> definition to express the fact that LexicalConcepts provide meaning to
> existing Lexical Entries.
>
>
>
> Cheers,
>
>
>
> Armando
>
>
>
> P.S I might be not recalling some other discussion or the rationale for
> which LexicalConcept was changed to be this way, but I wondered it could
> just have been an inexact definition given a posteriori when we were
> writing the specs.
>
>
>
>
>
> [1] George A. Miller and Richard Beckwith and Christiane Fellbaum and
> Derek Gross and Katherine Miller. Introduction to WordNet: An On-line
> Lexical Database. International Journal of Lexicography, 1990.
>
Received on Thursday, 20 April 2017 10:56:50 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 16:36:57 UTC