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

Re: Minutes teleconference last Friday

From: Philipp Cimiano <cimiano@cit-ec.uni-bielefeld.de>
Date: Thu, 21 May 2015 22:39:21 +0200
Message-ID: <555E4279.8030109@cit-ec.uni-bielefeld.de>
To: public-ontolex@w3.org
Hi Elena, see below....

Am 13.05.15 um 17:43 schrieb Elena Montiel Ponsoda:
> Dear Philipp,
>
> Some comments (to format, structure…), typos, and reflections to the 
> Core Module part:
>
> 1.In our (humble) opinion, we should check that all definitions follow 
> the same format, don’t you think so? See for example, ObjectProperty: 
> Usage and ObjectProperty Domain. The former starts with “The object 
> property…” and the latter “Provides…”.
>

         Yes, of course, I have been trying my best to homogeneize all 
the definitions. I am simply not managing to keep up with newly 
introduced properties...

> 2.The other of items (Domain, Range, Characteristics…) is not 
> consistent in all classes, i.e. it does not keep parallelism. For 
> instance, in the Object properties Sense and Reference, the order is 
> not the same.
>
               OK, I can do this at the end.

> 3.In the Object Property Sense there is a typo in the word 
> Characteristics.
>
     Done

> 4.In the class LexicalSense there is a typo in “it might nottttttttt 
> be directly”.
>
     Done.


> 5.In the Usage object property the preposition “of” is missing… “in 
> the use oofff”.
>
           Done.

> 6.Also, the sentence just before example 12 has some mistakes: “The 
> details of conditions are mostly left to application IN (preposition 
> missing!) specific vocabularies, but they could be given as natural 
> _langauge_ descriptions, e.g.,”.
>

          Is it better now?

> 7.Shouldn’t we specify a Range for the Usage object property?
>
          Yes, I specified it to be rdfs:Resource, is this fine?

> 8.What is the difference between rdfs:Literal and rdf:StringLang?
>
The class |rdf:langString| is the class of language-tagged string values 
<http://www.w3.org/TR/rdf11-concepts/#dfn-language-tagged-string>.

          The class |rdfs:Literal| is the class of literal 
<http://www.w3.org/TR/rdf11-concepts/#section-Graph-Literal> values such 
as strings and integers.

          So my understanding is that rdf:langString has to be 
language-tagged, while literals do not have to be tagged.

> 9.In example 15, we would say that a more appropriate lexicalization 
> of the spouse relation in Spanish would be “casarse” or even “casarse 
> con”, since it is a reflexive use of the verb.
>
               Yes, but the lemma canonical form is still "casar". The 
fact that casar is reflexive and subcategorizes a prepositional object 
introduced by the preposition "con" is something for the synsem module.

             Are you suggesting to change the canonical form or the URIs 
of the lexical entry?


> 10.In the ObjectProperty: Evokes, you refer to the propertyChain sense 
> or isLexicalizedSense of, what is the propertyChain doing?
>

         This is symmetric to the property chain we introduced for 
"denotes". denotes was introduced as a shortcut for a chain of sense o 
reference. So the axiom is saying that \forall x,y,z sense(x,y) \wedge 
reference(y,z) -> denotes(x,z)

         For evokes we have analogously:


         \forall x,y,z sense(x,y) \wedge isLexicalizedSenseOf(y,z) -> 
evokes(x,y)

         In words: if x has sense y and y is a possible lexicalization 
of a concept y, then x evokes the concept z.

        Clearer now?
>
> 11.In example 18, is there a reason for indistinctly using 
> “ontolex:isConceptOf dbpedia:Tuberculosis ;” and “ontolex:isConceptOf 
> <http://dbpedia.org/resource/Consumption_(Economics)> ;”.
>

          No, just sloppyness, I corrected this....

Thanks for all the feedback....

Philipp.
>
> We are working on the SynSem module. Comments to come next week. :)
> Shönen Himmelfahrt Tag!
> Best,
> Lupe and Elena
>
>
> El 11/05/2015 a las 9:04, Philipp Cimiano escribió:
>> Dear all,
>>
>>  I summarize the outcomes and decisions made during the telco last 
>> Friday:
>>
>> 1) Domain of ontolex:language
>>
>> We decided to unconstrain the domain of ontolex:language and instead 
>> add a = 1 ontolex:language axiom to the ontolex:Lexicon, 
>> ontolex:LexicalEntry and lime:LexicalizationSet classes. The doman of 
>> ontolex:language would thus be OWL:thing
>>
>> 2) Form should have minimum one writtenRep with range rdf:langString
>>
>> 3) The range of ontolex:language should be rdfs:Literal
>>
>> 4) We should drop the constraint on there being just one written 
>> representation per language tag as this is questionable and further 
>> it can not be axiomtaized in OWL anyway.
>>
>> 5) Example 10 is infelicitous as there should be two lexical entries 
>> for "bank" as in the case of geographic vs. financial meaning this is 
>> a case of homonymy. So in this case there should be two lexical 
>> entries. I will correct the example.
>>
>> 6)  Example 9: for the sake of this example I will change the 
>> denotation to <http://dbpedia.org/resource/Marriage>
>>
>> 7) We decided to keep dct:subject as property to assign a topic to a 
>> sense as this is the corresponding property from Dublin Core for this:
>>
>> 8) We decided to change the definition of affix to:
>>
>> "The class affix represents is a morpheme (suffix, prefix, infix, 
>> circumfix, etc.) that is attached to a word stem to form a new word."
>>
>> 9) We decided to change the definition of lexical entry to:
>>
>> "Lexical Entry is a unit of analysis of lexicon, that consist of a 
>> set of forms that are grammatically related and a set of base 
>> meanings that are associated with all of these forms. Thus, a lexical 
>> entry is a word, multiword expression or affix with a single 
>> part-of-speech, morphological pattern, etymology and set of senses."
>>
>> 10) We agreed to have domain and ranges for all properties, so I will 
>> add also owl:Thing to the range of "reference" for the sake of 
>> completeness (we discussed this differently during the telco, but to 
>> ensure consistency I propose we indeed explicitly indicate the range 
>> here, same for isSenseOf).
>>
>> 11) We decided to rename the property "condition" to "usage" and add 
>> it to the core module.
>>
>> 12) We briefly repeated the rationale for declaring a Lexicon as a 
>> dataset.
>>
>> TODOs:
>>
>> Elena/Lupe: to send me the updated definition of "Lexical Sense"
>> John: fix the namespaces
>>
>> That's it for now. Thanks to all those who attended the telco.
>>
>> I will implement these changes today.
>>
>> The next telco will be on the 22nd of Mai, 16:00 CET. We will discuss 
>> the synsem module then.
>>
>> I will send an email on this soon.
>>
>> Kind 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
Received on Thursday, 21 May 2015 20:39:54 UTC

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