Re: SSN Ontology Recommendation: inquiry about SSNX satisfiability

Dear all,

First, the online version of ssn-dul at http://www.w3.org/ns/ssn/dul does
not correspond to the latest one described in the spec. It should
correspond to the one at
https://github.com/w3c/sdw/tree/gh-pages/ssn/integrated/

I think that's a minor issue and can be fixed really fast by François.
(@François, I'll send you the RDF/XML version as well)
I reported this as ISSUE #997 on github.


Still, there is an insatisfiability whose origin is hasResult.

In a nutshell, the axioms (1) and (2) are too strict and should be relaxed
to (1') and (2') :

sosa:hasResult ≡ oldssn:hasValue (1)
sosa:hasResult ≡ oldssn:observationResult (2)

oldssn:hasValue ⊑ sosa:hasResult (1')
oldssn:observationResult ⊑ sosa:hasResult (2')


Axiom (1) is causing the following problem.
oldssn:hasValue is a sub-property of dul:hasRegion, whose range in DUL has
dul:Abstract as a super-class.
Yet, a sosa:Sampling procedure have only sosa:Sample as a result, and
sosa:Sample is a sub-class of ( dul:Event or dul:Object or dul:Quality )

Axiom (2) is causing a few problem that are non trivial, but basically it
boils down to either:
 - disjunction between dul:SocialObject and dul:PhysicalObject .
 - disjunctions between dul:Event and dul:Object .

If one relaxes (1) and (2) into (1') and (2'), then the insatisfiability no
longer holds.


I reported this as ISSUE #998 on github so that other editors can help
making the appropriate decision here.
https://github.com/w3c/sdw/issues/998

Best,
Maxime

Le mer. 17 janv. 2018 à 00:00, Maxime Lefrançois <maxime.lefrancois@emse.fr>
a écrit :

> Hi Sergio, all,
> That's odd, let me check on this one tomorrow.
> I'll keep you updated
> Best,
> Maxime
>
> Le mar. 16 janv. 2018 à 23:20, Sergio José Rodríguez Méndez <
> srodriguez@pet.cs.nctu.edu.tw> a écrit :
>
>> Hi, Francois.
>>
>> Thanks for the update.
>>
>> I have found another issue: this time is related to the SSNX
>> satisfiability. I ran HermiT v1.3.8 reasoner from Protege to check the
>> results on SSNX <https://www.w3.org/TR/vocab-ssn/#SSNX_Alignment>. There
>> are some classes that are labeled as unsatisfiable:
>>
>> [image: Inline image 1]
>>
>> Should all classes be satisfiable?
>>
>> /$¡rm
>> Best regards,
>>
>> Sergio José Rodríguez Méndez (羅士豪)
>> Pervasive Embedded Technologies Laboratory (PET Lab)
>> Computer Science Department
>> National Chiao Tung University
>>
>> On Tue, Jan 16, 2018 at 6:04 PM, Francois Daoust <fd@w3.org> wrote:
>>
>>> Hi Sergio,
>>>
>>>
>>>
>>> Many thanks for reporting these errors!
>>>
>>>
>>>
>>> I created two issues on the group’s issue tracker to track resolution on
>>> the Editor’s Draft:
>>>
>>>    - Range of actsOnProperty and isActedOnBy still incorrect in SOSA
>>>    figure:
>>>    https://github.com/w3c/sdw/issues/995
>>>    - Apartment 134 example incorrectly includes a sosa:actuationMadeBy
>>>    property:
>>>    https://github.com/w3c/sdw/issues/996
>>>
>>>
>>>
>>> The first one had already been identified and fixed in the SOSA+SSN
>>> actuation figure, but:
>>>
>>>    1. The fix did not make it to the published Recommendation for some
>>>    reason (similar fixes on other figures that were done in the same commit
>>>    were included, so I really don’t understand how this can have happened, but
>>>    it did…)
>>>    2. The SOSA-only actuation figure has not been fixed.
>>>
>>>
>>>
>>> As usual, we cannot update the published Recommendation directly
>>> unfortunately. I will complete the list of known errata.
>>>
>>>
>>>
>>> Thanks,
>>>
>>> Francois.
>>>
>>
>>