W3C home > Mailing lists > Public > public-ontolex@w3.org > November 2013

Re: [public-ontolex] <none>

From: John P. McCrae <jmccrae@cit-ec.uni-bielefeld.de>
Date: Thu, 14 Nov 2013 09:40:10 -0500
Message-ID: <CAC5njqoRb8=G1K2dqnaugvyX17bbvU8PBLH0F5rO=hOKDSjKuA@mail.gmail.com>
To: Armando Stellato <stellato@info.uniroma2.it>
Cc: "public-ontolex@w3.org" <public-ontolex@w3.org>
Hi Armando,

I think the point I was trying to make was that the metadata module should
IMHO only cover metadata about ontology-lexica. As such the metadata module
should be more or less limited to describing resources, which are
representable by means of the OntoLex-Lemon model.

What puzzles then is that some of these categories seem very strange in
description of a static RDF document. In particular, I mean that the
concept of a dictionary being "directional" is odd in this case. i.e., if
we have a resource, which gives entries for an ontological concept in two *or
more** languages then it is quite unclear in which direction the
"translation" occurs. Also, it has been my assumption that if we do
introduce a *translation* property linking senses then this property would
be symmetric.

Perhaps, you could clarify what is meant by directionality in terms of the
lexicon-ontology interface?

* Out of interest why is there no metadata for a lexicon with more than 2
languages?

Regards,
John

PS will add some more minor comments to the wiki about the metadata module



On Thu, Nov 14, 2013 at 6:52 AM, Armando Stellato <stellato@info.uniroma2.it
> wrote:

> Dear all,
>
> one important question raised by John in the last meeting, about LIME
> (LInguistic MEtadata...or...the green brother of LEMON :-) ).
>
> "Ok, I understand the need for metadata about how an ontology is
> enriched/covered by lexical info, but is in the scope of Ontolex to provide
> metadata/categories about linguistic resources?" (John, I hope I
> synthetized your point appropriately)
>
> In:
> http://www.w3.org/community/ontolex/wiki/Metadata_Module
>
> I reported the first draft of LIME vocabulary.
>
> As you may see, there are properties/classes for synthetizing the
> "linguistic asset" of an ontology.
>
> Then, there are classes representing various kind of linguistic resources
> (Dictionary, BilingualDictionary) and their switchable characteristics
> (hasGlosses, hasTaxonomy etc..). And these were the focus of John's
> question.
>
> I give here a few reasons for which these should be part of Ontolex:
>
> 1) if we want to describe how ontologies can be interface with lexicons,
> we should be able to provide enough details about the lexicon (though this
> has not to include the description of a Linguistic Resource)
>
> 2) we already mapped WordNet, and I thought we were addressing how to map
> various LRs
>
> 3) we have a module about lexical linked data, isn't the above metadata
> about that lexical linked data which we already included in the scope?
> I then could provide other motivations for interesting scenarios: for
> instance, ten years ago, I wrote a (initially) simple plugin for Protege,
> called OntoLing, usable for enriching ontologies with content coming from
> linguistic resources. The plugin attracted the attention of many people who
> found finally a tool for performing this task, and the good thing is that
> the tool could be easily  customizable throug plugins. The theory behind
> that is: "tell me what a resource is, and I tell how to use it at its
> best". We thus could easily plug WordNet, EuroWordnet, FreeDict, Freelang
> dictionaries etc.. and we could add easily as well add other resources
> provided by partners in specific projects. This description later took a
> life on its own, being called Linguistic Watermark.
>
> Other important scenarios are ontology mapping ones: today after years of
> OAEI contests, still access to lexical resource for supporting matchhing,
> is considered foundamental, though it is still not:
> 1) clear
> 2) systematic
> and including the support from ling resources is more a craftwork than a
> clear asset in any alignment workflow/architecture
>
> The Linguistic Watermark was initially only a software engineering
> solution for modeling heterogeneous resources (not available as RDF) to be
> dynamically loaded into a consistent framework for enriching ontologies.
> Years later, with RDF technologies getting more ground, time was mature for
> thinking about a way to represent this info in RDF, that's were (part of)
> the LW became the ancestor of what we are now presenting as LIME.
>
> All of the above has surely something to do with interfacing Ontologies
> with Lexicons.
>
> Ok, obviously, I had to advocate the thing ;-) but I totally understand
> the importance of the question from John, so...discussion is open!
>
> Cheers,
>
> Armando
>
> P.S: This email was also agreed last week, to provide the needed
> background, in order to start the discussion in our meetings
>
>
>
> --
>
> ----------------------------------------------------------------------------
> Armando Stellato, PhD
> AI Research Group at Tor Vergata (ART)
> Dept. of Computer Science, Systems and Production
> University of Roma Tor Vergata
> Via del Politecnico 1
> 00133 Roma, Italy
> email: stellato@info.uniroma2.it
> home: http://art.uniroma2.it/stellato
> office +39-06-72597330
> lab.   +39-06-72597332
> fax    +39-06-72597460
>
> ----------------------------------------------------------------------------
>
Received on Thursday, 14 November 2013 14:40:42 UTC

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