Re: R: telco on Friday

Hi Dave,

Thanks for your comments. In fact my proposed example is wrong in two ways:
the "qualify" pattern should have been used for the "generation" relation,
and it is may better not to attach the confidence to the activity directly,
unless you are sure that all the produced translations have the same
confidence value (e.g., taking always 1.0 for human translation). I have
re-built the provenance example at
https://www.w3.org/community/ontolex/wiki/Final_Model_Specification#Translation
to illustrate how the same translation can be generated  as result of two
different activities with two different confidence degrees (following the
qualify pattern for "generation"). For instance,

:bench_banco-trans prov:qualifiedGeneration [
     a prov:Generation ;
     prov:activity :humanValidationActivity ;
     :translationConfidence 1.0 ;
] .

:bench_banco-trans prov:qualifiedGeneration [
     a prov:Generation ;
     prov:activity :myTranslationAlgorithm ;
     :translationConfidence 0.3 ;
] .

I am still not convinced with the result, though. Maybe it's simpler to
attach a "confidence" attribute to the translation directly, along with its
"dc:source" (so there would be a different instance of Translation per
dc:source) and leave only the PROV-O-based mechanism when an exhaustive
report on the provenance chain is required. We can discuss this during  the
next telco.

Best regards,
Jorge




2014-10-27 21:36 GMT+01:00 dave.lewis@cs.tcd.ie <dave.lewis@cs.tcd.ie>:

>  Hi Jorge, All,
> Tracking this remotely and sporadically, so apologies in advance if I
> raises issues already resolved.
>
>  In your example Jorge, you annotate the prov:activity with the confidence
> score:
>
> #PROVENANCE
> :tranSetEN-ES a prov:Entity .
> :ApertiumLMF_EN-ES a prov:Entity ;
>                    owl:sameAs <http://hdl.handle.net/10230/17110> <http://hdl.handle.net/10230/17110> .
> :ApertiumLMF_EN-ES_Creators a prov:Agent .
> :tranSetEN-ES prov:wasDerivedFrom :ApertiumLMF_EN-ES .
>
> :translation a prov:Activity ;
>      prov:generated :ApertiumLMF_EN-ES ;
>      prog:agent :ApertiumLMF_EN-ES_Creators ;
>      :hasConfidenceDegree 1.0 .
>
>
> However, does that not mean you have to have a separate activity for each
> sense translation, since each could have a separate confidence? This would
> cause issues in likely scenarios where the same agent translates many
> senses, each with a different confidence degree. Similar issues might arise
> if you have multiple Translations in your TranslationSet, each with
> different confidence degrees.
>
> In previous thread:
> http://lists.w3.org/Archives/Public/public-ontolex/2014May/0030.html
> Philip, correctly in my opinion, identified that the confidence value, in
> this case 'hasConfidenceDegree' should be an attribute of the association
> of between the translated senses, i.e. the vartran:Translation object, in
> this case :bench_banco-trans
>
> However, if I'm interpreting the example correctly, you want to keep the
> provenance data, including the confidence degree, somewhat separate from
> the transet. This makes sense from a data management point of view.
>
> One way to do that, again in reference to the above thread, is to capture
> the translation relationship as a
> http://www.w3.org/TR/prov-o/#wasDerivedFrom relationship between the
> target and source senses, and reify it with the
> http://www.w3.org/TR/prov-o/#Derivation, using the 'qualify' pattern used
> in prov-o. This could be linked to the transset using an owl:sameAs and
> this Derivation, rather than the prov:Activity, would have the
> 'hasConfidenceDegree'.
>
> happy to work more on an example if you think this is worth pursuing.
>
> cheers,
> Dave
>
>
>
>
>
>
> On 24/10/2014 06:58, Jorge Gracia wrote:
>
> Hi Philipp, all
>
>  My apologies, my meeting started later than expected and I could not
> escape for the telco.  I will continue elaborating the example I sent
> (actually, I've just spotted an error).
>
>  Regards,
> Jorge
>
> 2014-10-24 14:06 GMT+02:00 Jorge Gracia <jgracia@fi.upm.es>:
>
>> Hi all,
>>
>>  I added a (still unfinished!) example of the potential usage of PROV-O
>> for representing translation confidence in the wiki, at the bottom of the
>> "translation" section
>> https://www.w3.org/community/ontolex/wiki/Final_Model_Specification#Translation
>>
>>  Talk to you later,
>>
>>  Jorge
>>
>>
>>
>> 2014-10-24 13:00 GMT+02:00 Manuel Fiorelli <manuel.fiorelli@gmail.com>:
>>
>>> Dear Philipp
>>>
>>> I have no issue in anticipating the discussion on LIME.
>>> Il 24/ott/2014 09:13 "Armando Stellato" <stellato@info.uniroma2.it> ha
>>> scritto:
>>>
>>>   Dear Philipp,
>>>>
>>>>
>>>>
>>>> for me it would be more than perfect, as today it would have been
>>>> really hard to participate to the whole session due to an urgent deadline.
>>>> Knowing it in advance, I can participate at 15:00 and maybe bail out by
>>>> 15:30.
>>>>
>>>> Think no issue for Manuel as well.
>>>>
>>>> Cheers,
>>>>
>>>> Armando
>>>>
>>>>
>>>>
>>>> *Da:* Philipp Cimiano [mailto:cimiano@cit-ec.uni-bielefeld.de]
>>>> *Inviato:* giovedì 23 ottobre 2014 21:44
>>>> *A:* public-ontolex@w3.org
>>>> *Oggetto:* Re: telco on Friday
>>>>
>>>>
>>>>
>>>> Hi Jorge,
>>>>
>>>>  sure, we can move the point to 15:30 and start with the metadata
>>>> module, but for that I need input from Armando and Manuel. Let's see if I
>>>> get something by tomorrow. I will then send an invitation to the telco at
>>>> short notice.
>>>>
>>>> Best regards,
>>>>
>>>> Philipp.
>>>>
>>>> Am 22.10.14 14:45, schrieb Jorge Gracia:
>>>>
>>>> Dear Philipp, all,
>>>>
>>>> Would it be possible to move the translation metadata example as last
>>>> point in the agenda? This Firiday I will connect around half an hour later
>>>> due to an overlap with another meeting. (I'll try to  have the example
>>>> ready by then)
>>>>
>>>> Regards,
>>>> Jorge
>>>>
>>>> El 22/10/2014 14:04, "Philipp Cimiano" <cimiano@cit-ec.uni-bielefeld.de>
>>>> escribió:
>>>>
>>>> Dear all,
>>>>
>>>>  I propose we devote the telco on Friday to the metadata module; if
>>>> people contribute in time, I would like to discuss the following:
>>>>
>>>> 1) Metadata about translation to represent confidence, source, etc:
>>>> Jorge wanted to prepare an example
>>>>
>>>> 2) lime examples: Armando, Manuel: can you look in detail at my two
>>>> examples from last week and send modified examples around so that we can
>>>> discuss them?
>>>>
>>>> 3) variation and translation module: discuss input from Elena (if she
>>>> is present)
>>>>
>>>> Please let me know if you can prepare the exaples until Friday to
>>>> discuss them.
>>>>
>>>> Best regards,
>>>>
>>>> Philipp.
>>>>
>>>> --
>>>> --
>>>> Prof. Dr. Philipp Cimiano
>>>> AG Semantic Computing
>>>> Exzellenzcluster für Cognitive Interaction Technology (CITEC)
>>>> Universität Bielefeld
>>>>
>>>> Tel: +49 521 106 12249 <%2B49%20521%20106%2012249>
>>>> Fax: +49 521 106 6560 <%2B49%20521%20106%206560>
>>>> Mail: cimiano@cit-ec.uni-bielefeld.de
>>>>
>>>> Office CITEC-2.307
>>>> Universitätsstr. 21-25
>>>> 33615 Bielefeld, NRW
>>>> Germany
>>>>
>>>>
>>>>
>>>>  --
>>>>
>>>> --
>>>>
>>>> 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/
>>
>
>
>
>  --
> Jorge Gracia, PhD
> Ontology Engineering Group
> Artificial Intelligence Department
> Universidad Politécnica de Madrid
> http://jogracia.url.ph/web/
>
>
> --
> Director - Knowledge and Data Engineering Group
> The CNGL Centre for Global Intelligent Content
> School of Computer Science and Statistics
> Trinity College Dublin
>
>


-- 
Jorge Gracia, PhD
Ontology Engineering Group
Artificial Intelligence Department
Universidad Politécnica de Madrid
http://jogracia.url.ph/web/

Received on Tuesday, 28 October 2014 18:56:34 UTC