Priklad CSVW to RDF

Step-by-Step Mapping with Annotations

id,title,author,published_date,price
1,The Catcher in the Rye,J.D. Salinger,1951-07-16,8.99
2,To Kill a Mockingbird,Harper Lee,1960-07-11,7.99
3,1984,George Orwell,1949-06-08,6.99

Step 1: Define JSON-LD Metadata with Mappings

Here, we define the metadata, specifying how each column in the CSV is mapped to RDF properties.

{
  "@context": "http://www.w3.org/ns/csvw",
  "url": "books.csv",
  "tableSchema": {
    "columns": [
      {
        "name": "id",
        "titles": "id",
        "datatype": "integer",
        "propertyUrl": "http://example.org/schema/id",
        "valueUrl": "http://example.org/books/{id}"
      },
      {
        "name": "title",
        "titles": "title",
        "datatype": "string",
        "propertyUrl": "http://schema.org/name"
      },
      {
        "name": "author",
        "titles": "author",
        "datatype": "string",
        "propertyUrl": "http://schema.org/author"
      },
      {
        "name": "published_date",
        "titles": "published_date",
        "datatype": "date",
        "propertyUrl": "http://schema.org/datePublished"
      },
      {
        "name": "price",
        "titles": "price",
        "datatype": "decimal",
        "propertyUrl": "http://schema.org/price"
      }
    ],
    "primaryKey": "id"
  },
  "rdfs:label": "Books",
  "rdfs:comment": "A list of books and their details",
  "dc:creator": "Tomas"
}

Annotations Explained:

1. @context: Defines the context for interpreting the metadata according to the CSVW standard.

2. url: The location of the CSV file to be mapped.

3. tableSchema: Describes the structure of the table, including the columns.

   Within columns:

   • name: Internal name for the column (matches the CSV header).
   • titles: The actual title as it appears in the CSV file.
   • datatype: Defines the data type of the column values (e.g., integer, string, date, decimal).
   • propertyUrl: Specifies the RDF property to which the CSV column is mapped. This URL represents the predicate in the RDF triple.
   • valueUrl (for id): Constructs a URI for each book, using the id value from the CSV. This URL will be used as the subject in the RDF triple.

4. primaryKey: Identifies the column that acts as the unique identifier for each row, ensuring each RDF subject is unique.

Step 2: Transformation into RDF

With the metadata defined, the transformation process maps each row in the CSV to RDF triples. Let's see how this looks for the first row:

@prefix ex: <http://example.org/books/> .
@prefix schema: <http://schema.org/> .

ex:1 a schema:Book ;
  schema:name "The Catcher in the Rye" ;
  schema:author "J.D. Salinger" ;
  schema:datePublished "1951-07-16"^^xsd:date ;
  schema:price "8.99"^^xsd:decimal .

Breaking Down the RDF Output:

1. Subject: ex:1 (constructed using the valueUrl from the id column). It uniquely identifies the book using its id.

2. Predicates and Objects:

   • schema:name maps to the title column and provides the book's title.
   • schema:author maps to the author column and specifies the author.
   • schema:datePublished maps to the published_date column and provides the publication date.
   • schema:price maps to the price column and specifies the book's price.

Key Points to Highlight:

• propertyUrl: This annotation is crucial as it directly maps CSV columns to RDF predicates. Each column in your CSV corresponds to a specific RDF property, ensuring that your CSV data can be accurately represented in RDF format.
• valueUrl: Used primarily for identifiers, this allows you to create unique URIs for your RDF subjects using the values in your CSV, ensuring that each resource is distinctly addressable on the web.

Summary

The process of mapping CSV columns to RDF involves defining a detailed metadata file that specifies how each piece of data in the CSV is to be interpreted and transformed into RDF triples. The propertyUrl and valueUrl are key annotations in the metadata that determine how the CSV data is converted into RDF, enabling semantic interoperability and ensuring that the data is machine-readable and linked on the web.