Add example mixing use of rml-views and gather maps #30

Activity

1. dachafra

   added

   blocked

   and removed

   blocked

   on Jan 18, 2023
2. frmichel

   added

   pendingIssue is pending due e.g. to a dependency with another or only applicable to a further releaseIssue is pending due e.g. to a dependency with another or only applicable to a further release

   documentationImprovements or additions to documentationImprovements or additions to documentation

   on Jan 18, 2023
3. frmichel commented on Sep 11, 2023

   

   frmichel

   on Sep 11, 2023

   Last edited by frmichel

   CollaboratorAuthor

   More actions

   @dachafra @chrdebru @andimou : To finalize this example I need the spec of the Fields (rml:field... ) but I fail to find it... Shouldn't it be in rml-io? Did i miss something?
4. dachafra commented on Sep 11, 2023

   

   dachafra

   on Sep 11, 2023

   Member

   More actions

   Fields is still in discussion. Indeed, in the joins task force, we are going to present the solution again and see if it covers all the requirements that we need. In any case, the issues that are labeled with pending do not need to be fixed now, as others need to be resolved before (as in this one)!
5. frmichel commented on Dec 15, 2023

   

   frmichel

   on Dec 15, 2023

   CollaboratorAuthor

   More actions

   This question was discussed during the meeting in Santiago de Compostela. Here is the solution that was agreed:
   @dachafra @chrdebru @andimou @bjdmeest @pmaria: could you please confirm that we understood the same thing?

   Input document "source.json":

   { 
     "id": 1,
     "a" : [ [1,2,3], [4,5,6] ]
   }

   The expected output is to generate a list of lists, where the head node of the outer list has a URI:

   <http://my.list/1>
     rdf:first (1 2 3);
     rdf:rest [ 
       rdf:first (4 5 6);
       rdf:rest rdf:nil 
     ].
   

   The solution involves a logical source for the json file, and a logical view that declares fields:

   <LS> 
       a rml:LogicalSource ;
       rml:source "source.json" ;
       rml:referenceFormulation ql:JSONPath;
   ].
   
   <LV>
       a rml:LogicalView ;
       rml:logicalSource <LS> ;
       rml:field [ 
         rml:fieldName "id" ;
         rml:reference "$.id" ;
       ] ;
       rml:field [ 
         rml:fieldName "a_string" ;
         rml:reference "$.a.*" ;
         rml:field [ 
           rml:fieldName "a_list" ;
           # Reference "$.*" applies to the results produced by evaluating "$.a.*"
           rml:reference "$.*" ;
           rml:groupBy "a"
         ]
       ]
   ].

   This reshapes the input document into a table with 2 lines:

   id	a_string	a_list
   1	[1,2,3]	1,2,3
   1	[4,5,6]	4,5,6

   Here, a_string is a string representation of the json array: "[1,2,3]", "[4,5,6]", so not really usable unless it would be parsed by a function for instance.
   Conversely, a_list is built by parsing the json arrays returned by "$.a.*", so it is an enumeration of the terms 1, 2, 3, and 4, 5, 6.
   The datatype of that list field has not been specified yet and requires further discussions.

   Now applying the following term map should generate the expected result:

   rr:subjectMap [ 
     rr:template "http://my.list/{id}";
     rml:gather ( [ 
       rml:gather ( [
         # "a_list" is multi-valued, it evaluates to the 3 terms  in each arrays: first iteration: 1, 2, 3, then 4, 5, 6.
         rml:reference "a_list" ;
         rml:gatherAs rdf:List;
       ] )
     ] );
     rml:gatherAs rdf:List;
   ] ;

   👍React with 👍2Reacted by Pano Maria and David Chaves
6. pmaria commented on Dec 17, 2023

   

   pmaria

   on Dec 17, 2023

   More actions

   Hi @frmichel,

   Indeed! Only the object of the rml:groupBy statement would be a_string, but I think that was your intention.

   So the logical view would be defined as such:

   <LV>
       a rml:LogicalView ;
       rml:logicalSource <LS> ;
       rml:field [ 
         rml:fieldName "id" ;
         rml:reference "$.id" ;
       ] ;
       rml:field [ 
         rml:fieldName "a_string" ;
         rml:reference "$.a.*" ;
         rml:field [ 
           rml:fieldName "a_list" ;
           # Reference "$.*" applies to the results produced by evaluating "$.a.*"
           rml:reference "$.*" ;
           rml:groupBy "a_string"
         ]
       ]
   ].
7. frmichel commented on Dec 17, 2023

   

   frmichel

   on Dec 17, 2023

   CollaboratorAuthor

   More actions

   Yes you're right, sorry for the typo. I renamed a and a2 into a_string and a_list for clarification, but I forgot to change a in groupBy ;).
8. dachafra

   added

   proposal

   and removed

   pendingIssue is pending due e.g. to a dependency with another or only applicable to a further releaseIssue is pending due e.g. to a dependency with another or only applicable to a further release

   on Dec 20, 2023
9. dachafra commented on Dec 20, 2023

   

   dachafra

   on Dec 20, 2023

   Member

   More actions

   Ok for me, I've also changed the label from pending to proposal
10. dachafra

    changed the title [-]Add example mixing use of fields and gather maps[/-] [+]Add example mixing use of rml-views and gather maps[/+] on Nov 6, 2024