tabulartordf.adoc

Tabular data to RDF data set

Use case

• A lot of data is tabular

• Timbuctoo is RDF store

• Map the tabular data to RDF

1. Upload tabular data

curl -v -F "file=@2017_09_28_BIA_Boerhaave_DEF.xlsx;type=application/vnd.openxmlformats-officedocument.spreadsheetml.sheet" -F "encoding=UTF-8" -H "Authorization: fake" http://localhost:8080/v5/u33707283d426f900d4d33707283d426f900d4d0d/mappingdemo/upload/table?forceCreation=true

33707283d426f900d4d33707283d426f900d4d0d is the (default) user in the local hosted Timbuctoo. For getting your user information see: Querying for (logged-in) User information mappingdemo is de name of the dataset.

Expected format

Excel

Header 1	Header 2	Header 3
Value 1	Value 2	Value 3
Value 1	Value 2	Value 3

CSV

Header 1;Header 2;Header 3
Value 1;Value 2;Value 3
Value 1;Value 2;Value 3

Supported types

• text/csv

• application/vnd.openxmlformats-officedocument.spreadsheetml.sheet (Excel spreadsheets)

• application/x.dataperfec (data perfect zips)

• application/x-msaccess (access databases)

2. Mapping the data

First make sure to create the right mapping file the language is RML in the JSON-LD format:

{
  "@context": {
    "rr": "http://www.w3.org/ns/r2rml#",
    "rml": "http://semweb.mmlab.be/ns/rml#",
    "tim": "http://timbuctoo.huygens.knaw.nl/mapping#"
  },
  "@graph": [
    {
      "@id": "http://timbuctoo.com/mappings/bia/plaatsnamen/Places",
      "rml:logicalSource": {
        "rml:source": {
          "tim:rawCollectionUri": {
            "@id": "http://example.org/datasets/u33707283d426f900d4d33707283d426f900d4d0d/bia/rawData/f70500b8-d1aa-41d3-850e-c5024f592857-file/collections/1"
          },
          "tim:customField": []
        }
      },
      "rr:subjectMap": {
        "rr:template": "http://timbuctoo.huygens.knaw.nl/v5/data/bia/Places/{persistent_id}",
        "rr:class": {
          "@id": "http://timbuctoo.huygens.knaw.nl/v5/data/bia/Places"
        }
      },
      "rr:predicateObjectMap": [
        {
          "rr:predicate": {
            "@id": "http://schema.org/name"
          },
          "rr:objectMap": {
            "rr:column": "name",
            "rr:datatype": {
              "@id": "http://www.w3.org/2001/XMLSchema#string"
            },
            "rr:termType": {
              "@id": "rr:Literal"
            }
          }
        },
        {
          "rr:predicate": {
            "@id": "http://schema.org/birthPlace"
          },
          "rr:objectMap": {
            "rr:parentTriplesMap": {
               "@id":"http://timbuctoo.com/mappings/bia/plaatsnamen/Places"
             },
             "rr:joinCondition": {
                "rr:child": "birth_place_persistent_id",
                "rr:parent": "persistent_id"
              }
          }
        }
      ]
    }
  ]
}

• http://timbuctoo.com/mappings/bia/plaatsnamen/Places is the name of the type.

• http://example.org/datasets/u33707283d426f900d4d33707283d426f900d4d0d/bia/rawData/f70500b8-d1aa-41d3-850e-c5024f592857-file/collections/1 is the name of the raw collection currently in Timbuctoo. The number the at the end of the URI is the tab number of the excel sheet. 1 means the fist sheet.

• rr:template in rr:subjectMap defines what the URI of the entity will be. This uri can be linked to from other types in this mapping.

• rr:predicate in rr:predicateObjectMap defines the rdf predicate will be used for column.

• rr:column uses the header (first line of the tabular file) of the column

• rr:joinCondition makes it possible to link between to resources

  • rr:child is the column that links to another resource (in SQL the foreign key field)

  • rr:parent is the id field in the resource that is linked (in SQL the primary key field)

You van use the following GraphQL query to retrieve it:

query metadata {
  dataSets {
    u33707283d426f900d4d33707283d426f900d4d0d__mappingdemo {
      metadata {
        collectionList {
          items {
            uri
          }
        }
      }
    }
  }
}

u33707283d426f900d4d33707283d426f900d4d0d__mappingdemo is the GraphQL id of the data set.

Note	Only use hte collections that contain rawData in the uri and that do not end with type.

When the mapping file is created the following curl command will execute the mapping:

curl --header "Content-type: application/ld+json" --header "Authorization: fake" --data "@{name of mapping file}.json" http://localhost:8080/v5/u33707283d426f900d4d33707283d426f900d4d0d/mappingdemo/rml

You can find the