Skip to content

semantic-columns.md

Latest commit

 

History

History
79 lines (53 loc) · 2.9 KB

semantic-columns.md

File metadata and controls

79 lines (53 loc) · 2.9 KB
Raw

Tutorial: describe columns semantically

Outcome

You will turn an ambiguous header such as temp into a column description with a human-readable explanation, a reusable semantic concept, and a machine-readable unit.

Before you begin

Load a table and connect Tabular file → Column Descriptions. Confirm that the expected headers appear.

1. Write a useful description

Describe what the value represents in this dataset. Prefer:

Ambient air temperature measured at the inlet sensor.

over:

Temperature column.

The first version gives a future reader context that is not present in the header.

2. Add a terminology concept

  1. Add Terminology Service to the canvas.
  2. Search for the scientific concept represented by the field.
  3. Review the result's label, definition, source ontology, and persistent IRI.
  4. Drag the best result into the column's description field.

Alternatively, connect Terminology Service → Column Descriptions. Each description field then searches concepts while you type; choose a suggestion to store its label and persistent IRI without dragging. Dragging still works.

Tabular displays the readable label and retains the IRI. Hover over the linked value to inspect it. Choose a concept because its definition matches the data, not merely because its label resembles the header.

3. Select a quantity and unit

  1. Connect Quantity Kinds → Units.
  2. Search Quantity Kinds for temperature or another measured property.
  3. Select the intended quantity kind.
  4. Drag the correct result from Units into the column's Unit field.

This copies the selected unit into that field. Alternatively, connect Units → Column Descriptions: each unit field then searches QUDT units while you type, filtered by the quantity kind selected in the connected Quantity Kinds → Units pair. Dragging still works.

The QUDT URI distinguishes units that may have similar names or symbols. Do not add units to identifiers, categories, free text, or genuinely unitless values.

4. Check the result

Connect Column Descriptions → RO-Crate. Its Turtle preview should contain the described column and, where selected, the concept and unit identifiers. Column descriptions are updated automatically; they have no separate Save button.

How Tabular represents columns

Tabular uses established vocabularies rather than a private column ontology:

  • W3C CSVW propertyUrl represents the semantic property expressed by a column, independently of its unit;
  • Dublin Core Terms represents textual descriptions; and
  • SDMX unitMeasure links the column to a controlled QUDT unit resource, following the CSVW Primer's fixed-unit-per-column pattern.

This RDF is complementary to the dataset metadata saved from Metadata Form. For the broader distinction, read Why Tabular.

Documentation guide · Next: Data and RO-Crate