Loading data using OntoRefine¶

Open OntoRefine in the Workbench¶

To transform your data into RDF, you need a working GraphDB database.

1. Start GraphDB in a workbench mode.

   graphdb
   

2. Open http://localhost:7200/ in a browser.

3. Go to Import -> Tabular (OntoRefine).

All data files in OntoRefine are organized as projects. One project can have more than one data file.

The Create Project action area consists of three tabs corresponding to the source of data. You can upload a file from your computer, specify the URL of a publicly accessible data, or paste data from the clipboard.

Create a project¶

1. Click Create Project -> Get data from.

2. Select one or more files to upload:

   • from your computer

     

   • from web addresses (URLs)

     

   • from clipboard

     

3. Click Next.

4. (Optional) Change the table configurations and update the preview.

   With the first opening of the file, OntoRefine tries to recognize the encoding of the text file and all delimiters.

   

5. Click Create Project.

   The result is table similar to that of an Excel or a Google sheet.

Import a project¶

To import an already existing OntoRefine project:

1. Go to Import Project.

2. Choose a file (.tar or .tar.gz)

3. Import it.

   

Open a project¶

Once the project is created

1. Go to Open Project.
2. Click the one you want to work on.
3. (Optional) Delete your project if you want to.

Using CONSTRUCT query in the OntoRefine SPARQL endpoint¶

Construct new RDF data based on the RDFized model of your tabular data. Use SPIN functions to construct proper triples.

GraphDB 8.O supports SPIN functions:

• SPARQL functions for splitting a string
• SPARQL functions for parsing dates
• SPARQL functions for encoding URIs

PREFIX mydata: <http://example.com/resource/>
PREFIX spif: <http://spinrdf.org/spif#>

# Example query returning RDF data
CONSTRUCT  {
  ?presidentIRI a mydata:President ;
        mydata:tookOffice ?tookOfficeParsed ;
        mydata:leftOffice ?leftOfficeParsed ;
        mydata:nominatedBy ?party ;
        mydata:homeState ?stateIRI
} WHERE {
    # Triple patterns for accessing each row and the columns in contains
    # Note that no triples will be generated for NULL values in the table
    # You should inspect your data in Refine mode and add OPTIONAL accordingly
    ?row a mydata:Row ;
        mydata:Presidency ?Presidency ;
        mydata:President ?President ;
        mydata:Wikipedia_Entry ?Wikipedia_Entry ;
        mydata:Took_office ?Took_office ;
        mydata:Left_office ?Left_office ;
        mydata:Party ?Party ;
        mydata:Portrait ?Portrait ;
        mydata:Thumbnail ?Thumbnail ;
        mydata:Home_State ?Home_State .

   # Uses SPIN function to parse the dates
    bind(spif:parseDate(?Took_office, "dd/MM/yyyy") as ?tookOfficeParsed)
    bind(spif:parseDate(?Left_office, "dd/MM/yyyy") as ?leftOfficeParsed)
    # Uses several functions to construct IRIs for the presidents and their states
    bind(iri(concat("http://example.com/", spif:encodeURL(?President))) as ?presidentIRI)
    bind(iri(concat("http://example.com/", spif:encodeURL(?Home_State))) as ?stateIRI)
} LIMIT 100

Records¶

For the purpose of the demo we can create records for each party. Go to Party column -> Edit column -> Move column to the beginning By Clicking Party column -> Text facet we can see that some values contain spaces.

1. Clean them by Party column -> Edit cells -> Transform and type value.trim(). Observe how the text facet is updated.

1. Now click Party column -> Sort -> OK Just use the defaults
2. And then Sort -> Reorder rows permanently

1. And Party column -> Edit cells -> Blank down. Click on Records. The result looks like this.

1. Now, the RDF button leads to the following query and results.

When there are records, the following URIs will be used in addition to the row ones:

• mydata:recordNNN - identifies a record, NNN is the 1-based record number;
• mydata:Record - the RDF type for each record;
• mydata:recordId - the property for record id (the value from the first column that creates an OpenRefine record);
• mydata:hasRow - the property that links records to rows.

Splitting content in a cell¶

Load the movie dataset for this example. Sometimes you may have multiple values in the same cell. You can easily split them and generate separate URIs using SPIF functions.

PREFIX mydata: <http://example.com/resource/>
PREFIX spif: <http://spinrdf.org/spif#>

SELECT * {
    ?row a mydata:Row ;
        mydata:genres ?genres ;
} LIMIT 100

PREFIX mydata: <http://example.com/resource/>
PREFIX spif: <http://spinrdf.org/spif#>

CONSTRUCT {
  ?m a mydata:Movie ;
    mydata:hasGenre ?genre
} WHERE {
  ?m a mydata:Row ;
     mydata:genres ?genres .
  ?genre spif:split (?genres "\\|")
} LIMIT 100