[Dbpedia-discussion] some invalid domain, range, subPropertyOf

2014-12-05 Thread Vladimir Alexiev
1) Consider http://mappings.dbpedia.org/index.php/OntologyProperty:FirstAscent:
it specifies `rdfs:domain Mountain, Volcano`. 

The author of that mapping probably thought this means that the property 
`firstAscent` should apply to `Mountain` or `Volcano`.
But by RDFS semantics, when you specify multiple classes as domain/range for a 
property, then every subject/object of a property is inferred to have all these 
classes. Eg in the above case, any subject will be inferred to be both 
`Mountain` and `Volcano`.

Furthermore, the ontology generator doesn't emit two classes, but one invalid 
class URI:
```
dbo:firstAscent rdfs:domain http://dbpedia.org/ontology/Mountain,Volcano ;
```

2) Some subProperty statements have an object spelt in Uppercase.
- in some cases this leads to a statement that does not connect to the intended 
property (in this case dbo:medalist), e.g.:
```
dbo:silverMedalist rdfs:subPropertyOf dbo:Medalist
```
- in other cases it leads to a statement which links to a class, which is a 
mistake
```
dbo:senator rdfs:subPropertyOf dbo:MemberOfParliament .
```

Out of 62 subProperty declarations, 20 have this problem:

dbo:bronzeMedalist rdfs:subPropertyOf dbo:Medalist .
dbo:codeLandRegistry rdfs:subPropertyOf dbo:Code .
dbo:codeMemorial rdfs:subPropertyOf dbo:Code .
dbo:distanceToCapital rdfs:subPropertyOf dbo:Distance .
dbo:dutchMIPCode rdfs:subPropertyOf dbo:Code .
dbo:goldMedalist rdfs:subPropertyOf dbo:Medalist .
dbo:iso6391Code rdfs:subPropertyOf dbo:LanguageCode .
dbo:iso6392Code rdfs:subPropertyOf dbo:LanguageCode .
dbo:iso6393Code rdfs:subPropertyOf dbo:LanguageCode .
dbo:musicalKey rdfs:subPropertyOf dbo:Type .
dbo:officialSchoolColour rdfs:subPropertyOf dbo:ColourName .
dbo:otherWins rdfs:subPropertyOf dbo:Wins .
dbo:politicGovernmentDepartment rdfs:subPropertyOf dbo:Department .
dbo:protectionStatus rdfs:subPropertyOf dbo:Status .
dbo:rankingWins rdfs:subPropertyOf dbo:Wins .
dbo:senator rdfs:subPropertyOf dbo:MemberOfParliament .
dbo:silCode rdfs:subPropertyOf dbo:LanguageCode .
dbo:silverMedalist rdfs:subPropertyOf dbo:Medalist .
dbo:subTribus rdfs:subPropertyOf dbo:Tribus .
dbo:superTribus rdfs:subPropertyOf dbo:Tribus .

3) I'm not sure whether it's a good idea to have classes and properties that 
have the same name, except capitalization.
The mapping wiki uppercases properties, e.g.
http://mappings.dbpedia.org/index.php/OntologyClass:MemberOfParliament 
http://mappings.dbpedia.org/index.php/OntologyProperty:MemberOfParliament 
So the difference between these two is lost on people.
There are 152 terms with duplicate names (see attachment).
But I don't imagine it's feasible to change all these now...

---

Posted as https://github.com/dbpedia/extraction-framework/issues/287 :

Would be nice if the mapping wiki and/or the ontology generator checks this:
- no commas in domain/range
- internal subPropertyOf are lowercase
- internal subPropertyOf and subClassOf actually exist


--
Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
from Actuate! Instantly Supercharge Your Business Reports and Dashboards
with Interactivity, Sharing, Native Excel Exports, App Integration  more
Get technology previously reserved for billion-dollar corporations, FREE
http://pubads.g.doubleclick.net/gampad/clk?id=164703151iu=/4140/ostg.clktrk
___
Dbpedia-discussion mailing list
Dbpedia-discussion@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion


Re: [Dbpedia-discussion] some invalid domain, range, subPropertyOf

2014-12-05 Thread Kingsley Idehen

On 12/5/14 11:19 AM, Vladimir Alexiev wrote:

1) Considerhttp://mappings.dbpedia.org/index.php/OntologyProperty:FirstAscent:
it specifies `rdfs:domain Mountain, Volcano`.

The author of that mapping probably thought this means that the property 
`firstAscent` should apply to `Mountain` or `Volcano`.
But by RDFS semantics, when you specify multiple classes as domain/range for a 
property, then every subject/object of a property is inferred to have all these 
classes. Eg in the above case, any subject will be inferred to be both 
`Mountain` and `Volcano`.


RDF Schema spec is clearly broken, in this regard.  Most would expect 
the domain of a property to indicate the nature of relation (represented 
by RDF statements) subjects, and the range to do the same for relations 
objects.


RDF Schema (as you've indicated, based on current spec [1])  asserts 
that given the following definition of the nature of a #related 
predicate (sentence/statement forming relation type):


#related rdfs:domain foaf:Person, foaf:Document .

the following entity description, represented by the #related relation 
below:


#this #related #that .

implies that:

#this is an instance of disjoint classes foaf:Person and 
foaf:Document, rather than the fact that the #related relation have 
subjects that are foaf:Person or foaf:Document class instances .


This is a nice example of why Schema.org introduced the 
schema:domainincludes [2] property.


Links:

[1] http://www.w3.org/TR/rdf-schema/#ch_domain
[2] http://schema.org/domainIncludes .

--
Regards,

Kingsley Idehen 
Founder  CEO
OpenLink Software
Company Web: http://www.openlinksw.com
Personal Weblog 1: http://kidehen.blogspot.com
Personal Weblog 2: http://www.openlinksw.com/blog/~kidehen
Twitter Profile: https://twitter.com/kidehen
Google+ Profile: https://plus.google.com/+KingsleyIdehen/about
LinkedIn Profile: http://www.linkedin.com/in/kidehen
Personal WebID: http://kingsley.idehen.net/dataspace/person/kidehen#this




smime.p7s
Description: S/MIME Cryptographic Signature
--
Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
from Actuate! Instantly Supercharge Your Business Reports and Dashboards
with Interactivity, Sharing, Native Excel Exports, App Integration  more
Get technology previously reserved for billion-dollar corporations, FREE
http://pubads.g.doubleclick.net/gampad/clk?id=164703151iu=/4140/ostg.clktrk___
Dbpedia-discussion mailing list
Dbpedia-discussion@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion