W3C home > Mailing lists > Public > public-rdf-shapes@w3.org > September 2016

Re: Validating instances based on other instances

From: Miika Alonen <miika.alonen@csc.fi>
Date: Tue, 27 Sep 2016 08:42:26 +0300 (EEST)
To: Dimitris Kontokostas <kontokostas@informatik.uni-leipzig.de>
Cc: "public-rdf-sha." <public-rdf-shapes@w3.org>
Message-ID: <1354264826.241388.1474954946093.JavaMail.zimbra@csc.fi>
I dont see how that would work. There can also be multiple Datasets and Documents are expected have references to non existing datasets eg: 

ex:Dataset1 ex:identifier "12.123/example.1234" . 
ex:Dataset2 ex:identifier "12.1234/example.2345" . 
ex:Dataset3 ex:identifier "12.123/example.4321" . 

ex:Document1 ex:related "12.123/example.1234" . 
ex:Document1 ex:related "20.234/undefined.2234" . 
ex:Document1 ex:related "null" . 

This again is example of "bad" data that could be cleaned - but there are many use cases where resources are "linked" using literal references. 

- Miika 

From: "Dimitris Kontokostas" <kontokostas@informatik.uni-leipzig.de> 
To: "Miika Alonen" <miika.alonen@csc.fi> 
Cc: "public-rdf-sha." <public-rdf-shapes@w3.org> 
Sent: Monday, 26 September, 2016 22:20:25 
Subject: Re: Validating instances based on other instances 

Hi Miika, 
a workaround could be something like 
ex:A a sh:Shape ; 
sh:targetObjectsOf ex:identifier ; 
sh:property [ 
sh:path [ sh:inversePath ex:related ] 
sh:minCount 1; 

On Mon, Sep 26, 2016 at 6:34 PM, Miika Alonen < miika.alonen@csc.fi > wrote: 

Hi all, 

Question about how to validate instance literals (or irirefs) based on other instance values. 

I have a use case where i have literals that should be same as some other literals, for example: 


ex:Dataset ex:identifier "doi:10.3972/example.0145.db" . 

ex:Document ex:related "doi:10.3972/example.0145.db" . 


ex:Dataset ex:identifier "doi:10.3972/example.0145.db" . 

ex:Document ex:related "doi:null" . 

Is there other ways to validate this than Sparql constraints. I know i could do something like: 

< http://example.com/ns# > sh:prefix "ex" . 

a sh:Shape ; 
sh:targetClass ex:DocumentType ; 
sh:sparql [ 
sh:message "Values must be dataset identifiers" ; 
sh:select """ 
SELECT $this ?value 
$this ex:related ?value . 
?identifier ex:identifier ?value . 
""" ; 
] . 

ETC ... but sparql seems bit overkill and it is hard to write. Is there or could there be other ways to produce these queries? For example property paths where first resource would be the starting point of the property path, eg: 

a sh:Shape . 
sh:targetClass ex:DocumentType ; 
sh:property [ 
sh:predicate ex:related ; 
sh:equals ( ex:DatasetType ex:identifier ); 
] . 

- Miika 

Dimitris Kontokostas 
Department of Computer Science, University of Leipzig & DBpedia Association 
Projects: http://dbpedia.org , http://rdfunit.aksw.org , http://aligned-project.eu 
Homepage: http://aksw.org/DimitrisKontokostas 
Research Group: AKSW/KILT http://aksw.org/Groups/KILT 
Received on Tuesday, 27 September 2016 05:43:03 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:02:44 UTC