Should all nodes in the RDF graph be typed?

[fernanev]

• [x] 1. BNF and PP Converters Should all blank nodes have RDF type? Some of them still doesn't have, for example:

mus:U17_has_opus_statement   [ cidoc-crm:P106_is_composed_of  "73" ;
                               cidoc-crm:P3_has_note          "Op. 73"
                             ] ;

• [x] 2. BNF Converter

F22_Self_Contained_Expression has two RDF types for example:

<http://data.doremus.org/Self_Contained_Expression/F22/7cf25xxx>
        a   frbroo:F22_Self-Contained_Expression , mus:Self_Contained_Expression ;

Should mus:... be removed?

• [x] 3. PP Converter When specifying the RDF type of Denomination Controlled Acces Point, we could do it more readable converting this:

<http://data.doremus.org/Denomination_Controlled_Access_Point/M18/08960xxx>
        a    <http://data.doremus.org/ontology#M18_Controlled_Access_Point_Denomination> ;

in this:

<http://data.doremus.org/Denomination_Controlled_Access_Point/M18/08960xx>
        a    mus:M18_Controlled_Access_Point_Denomination ;

[rtroncy]

Part 1: blank nodes should be avoided in the first place, and they MUST be typed, so indeed, a type should be added.

Part 2: this was clearly a mistake, thanks for the fix @pasqLisena

Part 3: I agree @fernanev

[pasqLisena]

Question related to Part 3.

Our namespace is

• http://data.doremus.org/ontology#, like in the Protege project

or

• http://data.doremus.org/ontology/, like exported currently by marc2rdf?

I introduced the problem 3 because I relied on the protege version, and I have not realized of the presence of the latter in the code.

[pierrechoffe]

@fernanev @pasqLisena FYI I created some new properties in order to make relationships clearer when we (erroneously) relied on class names, e.g we had:

mus:U17_has_opus_statement   [ cidoc-crm:P106_is_composed_of  "27" ;  
                               cidoc-crm:P106_is_composed_of  "2" ;  
                               cidoc-crm:P3_has_note  "Op. 27 no 2"
                             ] ;

now we would have:

mus:U17_has_opus_statement   [ mus:U42_has_opus_number  "27" ;  
                               mus:U43_has_opus_subnumber  "2" ;  
                               cidoc-crm:P3_has_note  "Op. 27 no 2"
                             ] ;

Same thing for catalogue statement. Should I update Protégé ?

[pasqLisena]

Should I update Protégé ?

Yes, please. Thank you.

[rtroncy]

We should privilege the # version for the ontology (not the / version). Can you either fix MARC2RDF or record a new issue?

[pasqLisena]

For Part 1, the type should be added in multiple parts of the code.

My next commit will fix it for keys and genres, but probably will be not enough. Please @fernanev can you list every blank node you find without type?

[pierrechoffe]

@pasqLisena @fernanev just updated the DOREMUS ontology on Protégé

[rtroncy]

@pierrechoffe Reminder, use the doremus-ontology repository for issues related to the ontology.

[fernanev]

Please @fernanev can you list every blank node you find without type?

Yes, no problem, I will do that list asap.