W3C-CSV-ON-THE-WEB

W3C CSVW CSV + JSON metadata Web data.

Definition

W3C CSVW 4 recommendations December 2015: (1) Model for Tabular Data and Metadata on the Web: conceptual model tables, columns, rows, cells. (2) Metadata Vocabulary for Tabular Data: JSON-LD vocabulary @context csvw:Schema, table, columns, datatypes, virtual columns, transformations. (3) Generating JSON from Tabular Data on the Web: algorithm transformation CSV+metadata -> JSON-LD or plain JSON. (4) Generating RDF from Tabular Data on the Web: algorithm transformation CSV+metadata -> RDF triples (NTriples, Turtle, RDF/XML). Sample: metadata.json defines tables array, columns with names + datatypes (number, string, date, boolean, URI), constraints (required, unique, pattern regex, minimum/maximum), virtual columns (computed), foreignKeys (cross-table). Adoption: open data portals (data.gov.uk, EU Open Data Portal, ONS Office for National Statistics UK), academic data sharing, modest production adoption.

Origin

W3C CSV on the Web Working Group founded 2014 ; 4 W3C recommendations published 17 December 2015 ; WG closed 2016 ; CSVW Community Group continues W3C maintenance.

Example in context

data.gov.uk publishes UK road accidents 2023 CSV dataset (~140000 rows) with CSVW metadata.json: columns definition (accident_id integer, date xsd:date, longitude decimal, latitude decimal, severity 1-3 enum, etc.), foreign keys to vehicles.csv ; automatic validation constraint engines (csvlint), RDF Linked Data publishing transformation.

Related terms

• W3C JSON-LD 1.1 — JSON-LD base.