Functional properties, SameAs inference, validation about owlsharp

[mdesalvo]

Dear Benoit,
yes this is the best channel to interact with the project in every way :)

The reasoner will be updated to address the functional property case you described: a functional property linking an individual to 2+ individuals can only mean that those individuals are the same. I'll introduce a fresh new standard rule to cover this.

The GlobalCardinality validator rule (which I understand is your actual pain point) will also be relaxed to not complain when detecting functional properties linking target individuals being declared or inferred the same.

Regards,
Marco

[benoitdaireaux]

Hi, Thanks for your swift response. I’ll monitor the progress to see when the changes are made. On a side note: at some point, we needed to parse some SPARQL query into a RDFSharp query. We couldn’t find such functionality in your library, so we started implementing a parser ourself. It is far from complete, and mainly parses SELECT queries, but please tell me if you’re interested in this type of code, I can point you to our repo. Regards, Benoit Daireaux Seniorforsker Senior Researcher [phone] [phone] Tullins gate 2, 0166 Oslo, Norway NORCE norceresearch.no<https://www.norceresearch.no/> From: Marco De Salvo ***@***.***> Sent: mandag 2. oktober 2023 18:30 To: mdesalvo/OWLSharp ***@***.***> Cc: Benoit Daireaux ***@***.***>; Author ***@***.***> Subject: Re: [mdesalvo/OWLSharp] Functional properties, SameAs inference, validation (Issue #7) Dear Benoit, yes this is the best channel to interact with the project in every way :) The reasoner will be updated to address the functional property case you described: a functional property linking an individual to 2+ individuals can only mean that those individuals are the same. I'll introduce a fresh new standard rule to cover this. The GlobalCardinality validator rule (which I understand is your actual pain point) will also be relaxed to not complain when detecting functional properties linking target individuals being declared or inferred the same. Regards, Marco — Reply to this email directly, view it on GitHub<#7 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ANDCMX56RMKTRVYGPFYP4EDX5LTY5AVCNFSM6AAAAAA5PTER4GVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTONBTGM2DAMBXGY>. You are receiving this because you authored the thread.Message ID: ***@***.******@***.***>>

[benoitdaireaux]

Hello Marco, I have some other comments. I added to my ontology a reasoner rule that states that under certain conditions to individuals should be considered the same (SameAs). I used first the OWLReasonerRuleSameAsAtom as a consequent in my rule. I noticed that the SameAs consequence was disregarded when the two individuals were already marked as DifferentFrom. Although I understand the logic, I have a scenario where I still would like to have this property generated, and then have the validator detect the inconsistency. It is perhaps a standard scenario, I am not sure what should be the expected behavior of the OWLReasonerRuleSameAsAtom in that case, but I just wanted to inform you. I worked around the problem by using an ObjectPropertyAtom with the OWL.SAME_AS property. However, I couldn’t find in the validation any check for individuals marked SameAs and DifferentFrom at the same time. So I decided to add my own validation rule. My problem now when writing the delegate that is passed as an argument to the OWLValidatorRule is that is must return an OWLValidationReport. The only constructor for such a OWLValidationReport is internal, and I can therefore not instanciate the report from my external code… Could the constructor be made public? I appreciate your help, Benoit Daireaux Seniorforsker Senior Researcher [phone] [phone] Tullins gate 2, 0166 Oslo, Norway NORCE norceresearch.no<https://www.norceresearch.no/> From: Marco De Salvo ***@***.***> Sent: mandag 2. oktober 2023 18:30 To: mdesalvo/OWLSharp ***@***.***> Cc: Benoit Daireaux ***@***.***>; Author ***@***.***> Subject: Re: [mdesalvo/OWLSharp] Functional properties, SameAs inference, validation (Issue #7) Dear Benoit, yes this is the best channel to interact with the project in every way :) The reasoner will be updated to address the functional property case you described: a functional property linking an individual to 2+ individuals can only mean that those individuals are the same. I'll introduce a fresh new standard rule to cover this. The GlobalCardinality validator rule (which I understand is your actual pain point) will also be relaxed to not complain when detecting functional properties linking target individuals being declared or inferred the same. Regards, Marco — Reply to this email directly, view it on GitHub<#7 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ANDCMX56RMKTRVYGPFYP4EDX5LTY5AVCNFSM6AAAAAA5PTER4GVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTONBTGM2DAMBXGY>. You are receiving this because you authored the thread.Message ID: ***@***.******@***.***>>

[mdesalvo]

Hi, Thanks for your swift response. I’ll monitor the progress to see when the changes are made. On a side note: at some point, we needed to parse some SPARQL query into a RDFSharp query. We couldn’t find such functionality in your library, so we started implementing a parser ourself. It is far from complete, and mainly parses SELECT queries, but please tell me if you’re interested in this type of code, I can point you to our repo. Regards, Benoit Daireaux Seniorforsker Senior Researcher [phone] [phone] Tullins gate 2, 0166 Oslo, Norway NORCE norceresearch.nohttps://www.norceresearch.no/

You are right, SPARQL parser is not available in the library since it would require unfeasible effort for the project. I would be grateful if you could contribute this as a feature though one or more PRs, even with initial scope set to SELECT queries only. Consider that we do not have a complete SPARQL algebra: a limited set of features (like MINUS operator and arithmetical property paths) are not implemented, so they should trigger error during the parsing process.

[mdesalvo]

Hello Marco, I have some other comments. I added to my ontology a reasoner rule that states that under certain conditions to individuals should be considered the same (SameAs). I used first the OWLReasonerRuleSameAsAtom as a consequent in my rule. I noticed that the SameAs consequence was disregarded when the two individuals were already marked as DifferentFrom. Although I understand the logic, I have a scenario where I still would like to have this property generated, and then have the validator detect the inconsistency. It is perhaps a standard scenario, I am not sure what should be the expected behavior of the OWLReasonerRuleSameAsAtom in that case, but I just wanted to inform you. I worked around the problem by using an ObjectPropertyAtom with the OWL.SAME_AS property. However, I couldn’t find in the validation any check for individuals marked SameAs and DifferentFrom at the same time. So I decided to add my own validation rule. My problem now when writing the delegate that is passed as an argument to the OWLValidatorRule is that is must return an OWLValidationReport. The only constructor for such a OWLValidationReport is internal, and I can therefore not instanciate the report from my external code… Could the constructor be made public? I appreciate your help, Benoit Daireaux Seniorforsker Senior Researcher [phone] [phone] Tullins gate 2, 0166 Oslo, Norway NORCE norceresearch.nohttps://www.norceresearch.no/

No problem, I'll open that ctor. It should be public of course.
Regarding the conflicts between sameAs and differentFrom they are checked at the moment of declaring them in the ontology data, so the ontology is always consistent. They are also checked in SWRL atoms, but I admit they cannot be checked outside these use cases at the moment. Maybe a standard validation rule for this would be appropriate.

[mdesalvo]

Three activities were done for this thread, now available in main after successful unit testing:

• Introduced standard reasoner rule OWLFunctionalEntailmentRule to emit owl:sameAs relations under situations in which functional object properties are found to link different target individuals and these individuals are compatible
• Modified standard validator rule OWLGlobalCardinalityRule to not complain when detecting functional property violations under situations in which target individuals are explicitly found to be related by owl:sameAs
• Modified visibility of OWLValidatorReport ctor in order to be public

[benoitdaireaux]

Thanks a lot! Benoit Daireaux Seniorforsker Senior Researcher [phone] [phone] Tullins gate 2, 0166 Oslo, Norway NORCE norceresearch.no<https://www.norceresearch.no/> From: Marco De Salvo ***@***.***> Sent: tirsdag 3. oktober 2023 20:18 To: mdesalvo/OWLSharp ***@***.***> Cc: Benoit Daireaux ***@***.***>; Author ***@***.***> Subject: Re: [mdesalvo/OWLSharp] Functional properties, SameAs inference, validation (Issue #7) Three activities were done for this thread, now available in main after successful unit testing: * Introduced standard reasoner rule OWLFunctionalEntailmentRule to emit owl:sameAs relations under situations in which functional object properties are found to link different target individuals and these individuals are compatible * Modified standard validator rule OWLGlobalCardinalityRule to not complain when detecting functional property violations under situations in which target individuals are explicitly found to be related by owl:sameAs * Modified visibility of OWLValidatorReport ctor in order to be public — Reply to this email directly, view it on GitHub<#7 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ANDCMX2BXQI34FKYSI67J3TX5RJGFAVCNFSM6AAAAAA5PTER4GVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTONBVGQ4TAMZVGI>. You are receiving this because you authored the thread.Message ID: ***@***.******@***.***>>

[mdesalvo]

@benoitdaireaux The library has been updated in Nuget with a new release (v3.8) containing these features.

Regards,
Marco