LanguageMap (or general literalelementmap): how to join?

[bjdmeest]

As it's a termMap, you can apply joins and data transformations similar to how other term maps work

What would a join on a language map look like?

Originally posted by @pmaria in #21 (comment)

Ah yes, I didn't think the joins too much through :P. This would require much more extensions, eg next to a TriplesMap also a LiteralElementMap that can have a different logicalSource, something like below.

@prefix rr: <http://www.w3.org/ns/r2rml#> .
@prefix foaf: <http://xmlns.com/foaf/0.1/> .
@prefix ex: <http://example.com/> .
@prefix xsd: <http://www.w3.org/2001/XMLSchema#> .
@prefix rml: <http://semweb.mmlab.be/ns/rml#> .
@prefix ql: <http://semweb.mmlab.be/ns/ql#> .
@base <http://example.com/base/> .

<TriplesMap1> a rr:TriplesMap;
    
  rml:logicalSource [ 
    rml:source "input-3/input.json";
    rml:referenceFormulation ql:JSONPath;
    rml:iterator "$.persons"
  ];

  rr:subjectMap [ 
    rr:template "http://example.com/persons/{firstname}";
  ];

  rr:predicateObjectMap [ 
    rr:predicate ex:lastName ;
    rr:objectMap [
      rml:reference "lastname";
      rml:languageMap [
        a rr:RefObjectMap ; # I guess we need a new name for this too, maybe a RefLiteralElementMap cfr. https://github.com/kg-construct/mapping-challenges/pull/21#issuecomment-738617830?
        rr:parentTriplesMap <TermMap1>;
        rr:joinCondition [
          rr:child "firstname" ;
          rr:parent "firstname" ;
        ]
      ]
    ]
  ].

## WARNING BELOW IS INTRODUCING WAAAY MORE CHANGES THAN ANTICIPATED

<TermMap1> a rr:LiteralElementMap;

  rml:logicalSource [ 
    rml:source "input-3/input.json";
    rml:referenceFormulation ql:JSONPath;
    rml:iterator "$.languages"
  ];

  rml:reference "language" .

I'm not sure we want to go this far.
This is (of course! :P ) also solvable using a fno:Function (as an fno:Function could refer to a different logicalSource), it would require something like below, but the whole 'using functions across different logicalSources' needs to be further specified

rml:languageMap [
  fnml:functionValue [
    rml:logicalSource [
      rml:source "input-3/input.json";
      rml:referenceFormulation ql:JSONPath;
      rml:iterator "$.languages"
    ];

    rr:predicateObjectMap [
      rr:predicate fno:executes ;
      rr:objectMap [ rr:constant idlab-fn:trueCondition; rr:termType rr:IRI ]
    ];

    rr:predicateObjectMap [
      rr:predicate idlab-fn:strBoolean ;
      rr:objectMap <#EqualsFunction>
    ];
			
    rr:predicateObjectMap [
      rr:predicate idlab-fn:str ;
      rr:objectMap [ rr:reference "language" ]
    ]
  ]
].

<#EqualsFunction>
	
  fnml:functionValue [
    rml:logicalSource [
      rml:source "input-3/input.json";
      rml:referenceFormulation ql:JSONPath;
      rml:iterator "$.persons"
    ];

    rr:predicateObjectMap [
      rr:predicate fno:executes ;
      rr:objectMap [ rr:constant idlab-fn:equal ]
    ];

    rr:predicateObjectMap [
      rr:predicate grel:valueParameter ;
      rr:objectMap [ rml:reference "languages" ] # TODO we need to specify which logical source we take this from
    ];
		
    rr:predicateObjectMap [
      rr:predicate grel:valueParameter2 ;
      rr:objectMap [ rml:reference "languages" ] # TODO we need to specify which logical we take this from
    ]
  ]
].

I have the feeling we could use xrr:pushDown here to join over different data sources as well (@frmichel what do you think?). We have options. But maybe the first question should be: do we want to go this complex? :P @VladimirAlexiev have you come across this use case?

[dachafra]

@bjdmeest which is the status of this? Is it still something we need to have? Or is it more like a "good-to-have" feature.

[bjdmeest]

I personally think this is solved once we have https://github.com/kg-construct/rml-fno-spec/issues/2 solved

[dachafra]

As this is already taken into account in the spec and we are working on it at https://github.com/kg-construct/rml-fno-spec/issues/2, we can close it