Generate owl:equivalentClass triples

[nleguillarme]

Hi,
As part of my knowledge graph creation process, I have some data in a csv file I need to transform into owl:equivalentClass triples.
My data basically consists of two columns, one with the subject IRI, the second with the object IRI, and I want to generate triples of the form (subject_iri, owl:equivalentClass, object_iri).
It does not seem possible, as mapeathor treats each subject as individuals. Here, my subjects are classes.
Am I right ?

[anaigmo]

Interesting point. I think yes, but by definition I think all mappings in [R2]RML create always subjects as instances. Please show an example if you have found a way to do it. I'm thinking, maybe, deleting the rr:class and in a POM stating that is an owl:class?

Anyway, this issue is more related to what RML and R2RML can represent. The only problem I can see that addresses Mapeathor if it forces the user to specify a rr:class for the subject. I meant, please check first if this can be done and correctly processed by an [R2]RML-processor, and if so, I'll see if it affects the current behaviour of Mapeathor, because right now I'm not sure if it is affected.

[nleguillarme]

Here is how I do it in YARRRML:

SRCTAXON:
    sources:
      - access: ./taxon.tsv
        referenceFormulation: csv
        delimiter: "\t"
    s: $(src_iri)
    po:
      - p: owl:equivalentClass
        o:
          mapping: TGTTAXON
          condition:
            function: equal
            parameters:
              - [str1, $(ID)]
              - [str2, $(ID)]

This generates the following RML rules

:rules_000 void:exampleResource :map_SRCTAXON_000.
:map_SRCTAXON_000 rml:logicalSource :source_004.
:source_004 a rml:LogicalSource;
    rml:source :csvw_004.
:csvw_004 a csvw:Table;
    csvw:url "./taxon.tsv";
    csvw:dialect :csvw-dialect_004.
:csvw-dialect_004 a csvw:Dialect;
    csvw:delimiter "\t".
:source_004 rml:referenceFormulation ql:CSV.
:map_SRCTAXON_000 a rr:TriplesMap;
    rdfs:label "SRCTAXON".
:s_004 a rr:SubjectMap.
:map_SRCTAXON_000 rr:subjectMap :s_004.
:s_004 rml:reference "src_iri".
:pom_010 a rr:PredicateObjectMap.
:map_SRCTAXON_000 rr:predicateObjectMap :pom_010.
:pm_010 a rr:PredicateMap.
:pom_010 rr:predicateMap :pm_010.
:pm_010 rr:constant owl:equivalentClass.
:pom_010 rr:objectMap :om_010.

[anaigmo]

I see, I've been running tests and mapeathor creates "rr:class nan" when provided with no class, so that should be changed in order to address your use case, the rest (I think) is fine. Tagging for modification

[anaigmo]

The new release v1.5.4 can create a subject with no class, I hope it works for you now!

[nleguillarme]

Hi @anaigmo, subject with no class creation works for me, thank you !
However, I still lack the possibility of specifying that the URI I read in my source file is a IRI -> the subject is recognized as a BlankNode. Would it be possible to add a "IsIRI" column in the Subject tab, so that if I set Class to empty and IsIRI to true, mapeathor recognized that my subject is a class ?

[anaigmo]

I'm glad that works :)

I'm re-opening the issue for the additional column issue. Could you provide an example of the IRI that is recognized as BN? So I can see why it behaving like this. I'd like to avoid adding more columns when possible, so first I want to check if it is possible to distinguish IRI/BN in all cases. Thanks!

[nleguillarme]

Here are a few examples

iri	queryIRI
https://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?id=288551	https://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?id=288551
http://www.gbif.org/species/2280552	https://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?id=288551

I would like to be able to create owl:equivalentClass triples between elements in the iri and queryIRI columns

[nleguillarme]

Hi, I just realized that what I meant all along was not to generate owl:sameAs triples but owl:equivalentClass triples (to indicate that two classes are equivalent). I have changed all my previous posts to reflect this.

I am still interested in creating triples relating classes (and not individuals).