W3C home > Mailing lists > Public > public-ontolex@w3.org > May 2015

Re: ontolex module ready for final discussion

From: Jorge Gracia <jgracia@fi.upm.es>
Date: Fri, 8 May 2015 12:29:54 +0200
Message-ID: <CANzuSaNgo40BHbrpfQHR0oDCyy6NHrAKHoJckZs53ciyL8MWKw@mail.gmail.com>
To: Aldo Gangemi <aldo.gangemi@cnr.it>
Cc: Philipp Cimiano <cimiano@cit-ec.uni-bielefeld.de>, Elena Montiel Ponsoda <emontiel@fi.upm.es>, "public-ontolex@w3.org" <public-ontolex@w3.org>
Dear all,

I agree with Aldo's proposal of using rdfs:Resource rather than owl:Thing
when we want to be unspecific about the type of range. owl:Thing would work
also, but I agree that the implications of punning might introduce
unnecessary extra complexity.

Regards,

Jorge


2015-05-08 12:17 GMT+02:00 Aldo Gangemi <aldo.gangemi@cnr.it>:

> Hi, not sure I can attend today’s telecon, some comments below.
>
> On May 7, 2015, at 10:52:20 PM , Philipp Cimiano <
> cimiano@cit-ec.uni-bielefeld.de> wrote:
>
> Hi Elena, all,
>
>  I try to answer your comments below...
>
> Regards,
>
> Philipp.
>
> Am 07.05.15 um 16:59 schrieb Elena Montiel Ponsoda:
>
> Dear Philipp, all
>
> We had read carefully the core module and would like to make the following
> comments:
> ·         Definition of LexicalEntry class: “The class *lexical entry* represents
> a single unit of analysis in the lexicon comprising a collection of forms
> that are morphologically related or have a single pronunciation and have
> the same set of meanings.”
> o   Why morphologically related “or” have a single pronunciation? Isn’t
> the use of “or” a bit inaccurate?
> o   Regarding the fact of having “a single pronunciation”, what happens
> with a word like “advertisement” with an American and a British
> pronunciation? Why should we have two lexical entries for this? In fact,
> the same lexical entry would be pointing to two different lexical forms
> with different phonetic representations, right?
>
>
>                 Well, exactly because of the "colour" example you mention
> below. The pronunciation of "colour" and "color" is the same, but they are
> not morphologically related, are they? We do not want to have two lexical
> entires, that's why we need the "OR" in the definition.
>
>                 For the case of "advertisement": it would be one entry
> with *one* form and two different pronunciations of this form. We can add
> this example if it help. Why are you implying that the definition requires
> to have two entries? It is one entry with one form and two pronunciations
> of the latter.
>
> ·         We would like to suggest this definition: The class lexical
> entry represents a single unit of analysis in the lexicon comprising a
> collection of forms with the same set of meanings and morphologically
> related.
>
>
>               see above
>
> ·         In the following sentence: “Lexical entries are further
> specialized into *Word*, *Affix* (e.g., suffix, prefix or circumfix) and
> *MultiwordExpression* for example”, we would suggest to include infix as
> an example of Affix and change “or” by “and”, since you are providing some
> examples.
> ·         As for the definition of Affix, we would suggest “The class
> *affix* represents is a morpheme (suffix, prefix, circumfix, etc.) that
> is attached to a word stem to form a new word.”, rephrasing Wikipedia… J
>
>
>              That's fine. I will adopt it.
>
> ·         Below example 2, you say: “one for the British English written
> representation "colour" and one for the American English written
> representation”, we would add “color”, i.e., the American Eng. Repr.
>
>
>               I do not understand, sorry.
>
> ·         General comment: The entities in ontolex.owl ontology do not
> have rdfs:label, is there a reason for that? In fact, we think there should
> be rdfs:labels in different languages.
>
>                Right, I will add the labels. Fortunately, this is not a
> principled problem ;-)
>
> ·         The explanation below example 4 “*Each form should have exactly
> one written representation per language tag*, but there are no
> restrictions on the number of associated phonetic representations”, could
> be debatable…  Why do we need to be so restrictive? “exactly one”?
>
>
>            Yes, this is something to talk about.
>
>
> Agree, it is better to relax this contraint, unless we can imagine
> reasoning advantages from the functionality deriving from it.
>
> ·         Below OtherForm: “For example we may specify non-canonical
> forms of the verb ""marry"" as follows:” Double quotation marks!
>
>
>            OK
>
> ·         Below example 10 : “Thus, it is not possible to specify the
> range of denotes in OWL”, we think that this statement is inaccurate or
> too categorical, since one could say owl:Thing. Moreover, in the definition
> of LexicalSense, the range as been defined as being owl:Thing
>
>
>               owl:Thing is the class comprising all owl:individuals, it
> does not include classes or properties... however using a property or class
> as reference makes an individual technically, so probably it is fine to add
> the axiom. In any, case owl:Thing is the default, so we do not need to add
> it.
>
>
> I think that for this generic denotation we need rdfs:Resource rather than
> owl:Thing. The literature is not keen about accepting logical primitive
> classes as subclasses of owl:Thing by punning, because that design creates
> complexity issues, even with punning.
>
>
> ·         Comment regarding the inclusion of properties in the core: why
> do we make explicit properties such as “morphologicalPattern” in the
> LexicalEntry class, and cannot include the property “usage” in the
> LexicalSense class? We think the inclusion of that property would make
> clearer what the purpose of having LexicalSense in the model is. For a
> complete categorization of the types of usages we would then be pointing to
> an external model…
>
>
>               We can.
>
> ·         As for the Definition property  you say that the domain can be
> a LexicalConcept or a LexicalSense, but this is neither reflected in the
> figure above nor in the LexicalSense and LexicalConcept descriptions.
>
>
>             I added this without consulting anybody else as I understood
> that was a request from the community. The picture has to be updated.
>
>
> definition has owl:Thing in the range, but this has issues because, if a
> definition is a gloss, we should have a Gloss or Definition class (while
> there is a datatype property gloss in wordnet schema). Anyway, if so, the
> range should be a Gloss/Definition class.
> Also: what if a named graph is provided as a definition? or any other
> structured form in a semantic lexicon? I’d extend the range to
> rdfs:Resource, in order to enable any URI to work as a definition
> identifier. We should also explain that literals are not acceptable.
> An alternative is to accept literals as values for a datatype property
> definition, but in that case, any kind of structured definition will be
> referenced as a literal, not by a URI.
>
> ·         Below the class LexicalSense: “Via the lexical sense object we
> can attach additional properties to a pair of lexical entry *and
> ontological predicate that it* denotes to describe under which conditions
> (context, register, domain, etc.) it is valid to regard the lexical entry
> as having the ontological entity as meaning. For example, we may wish to
> express the usages of the word "consumption" in terms of the topic and
> diachronic usage *of the word*.”: and the ontological predicate it
> denotes... in terms of the topic and its diachronic usage
>
>
>               What is the point here?
>
>
> I do not understand either
>
> ·         It is unclear why Lexicon is a subclass of void:Dataset (sorry
> if this has been discussed previously)
>
>
>              Yes, we discussed this to be able to attach metadata
> properties to a Lexicon as a dataset.
>
>
> I agree on this choice, if a dataset can be expressed in any data model or
> language.
>
> Aldo
>
> ·         In LexicalConcept: there is a spelling mistake in sublcas
> We hope they help!
>
> Elena, Lupe & Jorge
>
> El 04/05/2015 a las 22:19, Philipp Cimiano escribió:
>
> Dear all,
>
>  the core module is now ready for the final discussion this week. Please
> check the current version of the specification:
>
> https://www.w3.org/community/ontolex/wiki/Final_Model_Specification#Core
>
> The ontologies and code of examples is available here:
>
> https://github.com/cimiano/ontolex.git
>
> Please send me any final issues / changes you would like to have
> implemented in the specification by Thursday.
>
> Looking forward to our discussion on Friday!
>
> Access details are here, as usual:
> https://www.w3.org/community/ontolex/wiki/Teleconference,_2015.5.8,_16-17_pm_CET
>
> Best regards,
>
> Philipp.
>
>
>
> --
> --
> Prof. Dr. Philipp Cimiano
> AG Semantic Computing
> Exzellenzcluster für Cognitive Interaction Technology (CITEC)
> Universität Bielefeld
>
> Tel: +49 521 106 12249
> Fax: +49 521 106 6560
> Mail: cimiano@cit-ec.uni-bielefeld.de
>
> Office CITEC-2.307
> Universitätsstr. 21-25
> 33615 Bielefeld, NRW
> Germany
>
>
>


-- 
Jorge Gracia, PhD
Ontology Engineering Group
Artificial Intelligence Department
Universidad Politécnica de Madrid
http://jogracia.url.ph/web/
Received on Friday, 8 May 2015 10:30:42 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 16:36:49 UTC