Re: Semantic Predication: 1 - basic distinctions

> On 17. Feb 2023, at 17:22, Franconi Enrico <franconi@inf.unibz.it> wrote:
> 
>> On 17 Feb 2023, at 13:44, Thomas Lörtsch <tl@rat.io> wrote:
>> Am 17. Februar 2023 12:17:43 MEZ schrieb Franconi Enrico <franconi@inf.unibz.it>:
>>> The case of semantic predication stems from the Davidsonian event semantics tradition.
>>> Almost any verb in language induces an event. A sentence like ‘Liz married Richard’, together with being a statement (:liz :spouse :richard), induces the existence of an event e of type marriage marriage(e) with thematic roles agent(e)=Liz and patient(e)=Richard. When we say 'The marriage lasted from 1964 to 1974’, we add to the previously mentioned event (referred to by the anaphoric determiner ’The’) the thematic role period(e)=1964-1974. When we then say 'Liz married again Richard’ and ’This marriage lasted from 1975 to 1976’, we induce a distinct event e’ having different properties: marriage(e’), agent(e')=Liz, patient(e')=Richard, period(e')=1975-1976.
>>> 
>>> But in RDF, how do we say 'again’?
>>> My proposed ’trick’ is to create special types of marriage events, e.g., the first marriage, the second marriage, etc. Why does this trick work? Because in this case the two spouses together with the special type of marriage event (the first, the second, …) UNIQUELY identify all the other properties of any event instance in the domain of interest (the period, the place, the residence or whatever else).
>>> 
>>> The general principle to deal with multi-edges and semantic predication is to make sure while modelling that the event type together with the subject and the object uniquely identify any other property that the event may have. So, nothing to do with the Singleton Property.
>> 
>> In that case I fail to see the principle. Could you elaborate how a user is come to know that -1 refers to the first marriage, -2 to the second etc? Why not encode identification semantics in the same way, or referential transparency etc?
> 
> Nope. I don’t need to specify the meaning of spouse-1 and spouse-2 (both being subproperties of spouse); they just need to be two arbitrary subproperties of spouse. The graph nonambiguosly says that there are two special spouse predicates which, together with their subject and object, each one uniquely identifies an event of type marriage having each one different properties.
> There’s no need to specify more, unless you want to. For example, you can say - if you wish to convey this additional meaning - that spouse-1 and spouse-2 are the first-spouse and the second-spouse; I did this in the example above.

Okay, as I thought. In this case of course you don’t have to specify any naming semantics for those new properties.

>>  If I follow your argument then IMO what you call a trick is really just a hack, understandable only to you, and with no formal anything at all. 
> 
> No, the graph has exactly the needed formal semantics to express multi-edge cases. Everybody, given the semantics specified for RDF-star, would understand exactly what I said.

Yep.

>> Syntactically it still is a Singleton Property approach, and if you that accept it would be wise (and compositional) to add the extra meaning via separate annotations to the property instead of minting countless new properties that have special meanings baked into them, then it would also semantically be Singleton Properties. 
> 
> Well, as I have shown, this is not  a Singleton Property approach, and it does not require any extension to RDF-star — which gives the (various?) semantics to embedded triples.
> Nonetheless, I’m curious about understanding better the Singleton Property approach. I’m studying...

It seems to me that there’s no noteworthy difference between your "trick" and the Singleton Property approach. But I don’t want to rush you in your reading ;-)

Best,
Thomas

> cheers
> —e.
> 
> 
>>> On 16 Feb 2023, at 20:49, Olaf Hartig <olaf.hartig@liu.se> wrote:
>>> 
>>> Yes that's true  ...but only if Enrico's intention was indeed to use :spouse-1 and :spouse-2 as unique property names, which I am not sure of. Hence my question.
>>> 
>>> Olaf
>>> 
>>> 
>>> Feb 16, 2023 20:36:07 Souripriya Das <souripriya.das@oracle.com>:
>>> 
>>> The use of :spouse-1 and :spouse-2 as unique property names that are rdfs:subPropertyOf the main property :spouse is similar to one of the three approaches described in [1]. It is similar in idea to singleton properties but utilizes the preexisting property rdfs:subPropertyOf instead of requiring a new special property called rdf:singletonPropertyOf.
>>> 
>>> [1] https://eur03.safelinks.protection.outlook.com/?url=https%3A%2F%2Fopenproceedings.org%2FEDBT%2F2014%2Fedbticdt2014industrial_submission_28.pdf&data=05%7C01%7Cfranconi%40inf.unibz.it%7Cbd5a7c5945e24d67170908db10e4bde6%7C9251326703e3401a80d4c58ed6674e3b%7C0%7C0%7C638122347289371004%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=RboVfP0W9eU40MGxvedXus4iv%2F3sU5XBehd3Y66Tp5s%3D&reserved=0<https://eur03.safelinks.protection.outlook.com/?url=https%3A%2F%2Fopenproceedings.org%2FEDBT%2F2014%2Fedbticdt2014industrial_submission_28.pdf&data=05%7C01%7Cfranconi%40inf.unibz.it%7Cbd5a7c5945e24d67170908db10e4bde6%7C9251326703e3401a80d4c58ed6674e3b%7C0%7C0%7C638122347289371004%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=RboVfP0W9eU40MGxvedXus4iv%2F3sU5XBehd3Y66Tp5s%3D&reserved=0>
>>> ________________________________
>>> From: Olaf Hartig <olaf.hartig@liu.se>
>>> Sent: Thursday, February 16, 2023 1:37 PM
>>> To: tl@rat.io <tl@rat.io>; franconi@inf.unibz.it <franconi@inf.unibz.it>
>>> Cc: public-rdf-star-wg@w3.org <public-rdf-star-wg@w3.org>
>>> Subject: [External] : Re: Semantic Predication: 1 - basic distinctions
>>> 
>>> Hi Enrico,
>>> 
>>> Regarding Singleton Properties, I am not actually sure that your idea with the :spouse-1 and :spouse-2 properties in
>>> your example was the same as the idea of Singleton Properties (as Thomas' comment suggests it was). Or maybe it was?
>>> 
>>> To understand whether it was or not, let me ask you the following question. Was your intention with the :spouse-1
>>> property to represent a "first-spouse" relationship that can also be used between other couples? In other words, was
>>> your intended meaning of :spouse-1 such that, in addition to the triple (:liz, :spouse-1, :richard), there could also be
>>> a triple such as (:alice, :spouse-1, :bob)?
>>> 
>>> If that's the case, then this is something else than Singleton Properties.
>>> 
>>> Best,
>>> Olaf
>>> 
>>> 
>>> On tor, 2023-02-16 at 16:39 +0000, Franconi Enrico wrote:
>>>>> On 16 Feb 2023, at 16:01, Franconi Enrico <franconi@inf.unibz.it> wrote:
>>>>> 
>>>>>> In the last example on semantic predicatins in eMail nr. 2 you use properties ":spouse-1" and ":spouse-2", defined
>>>>>> as subproperties of ":spouse". Note that here you are employing the Singleton Property approach and wouldn't need
>>>>>> quoted triples at all. But, because quoted triples reference the type, practically all your examples could face
>>>>>> the same need to account for a multiplicity of annotations. Ergo Singleton Properties might be the better approach
>>>>>> after all.
>>>>> 
>>>>> I don’t know where to read in order to understand what the Singleton Property is (my fault, sorry…).
>>>> 
>>>> OK, I’m studying now the singleton property in Vinh Nguyen, Amit P. Sheth: Logical Inferences with Contexts of RDF
>>>> Triples (2017) [and previous references]; I’m not sure we need all that machinery, but some of the syntactic choices
>>>> are appealing.
>>>> I’l go deeper.
>>>> —e.
>>> 
>

Received on Wednesday, 22 February 2023 15:41:55 UTC