article

Overview

The article parent type covers authored content produced for publication. This is the most common parent type in production, representing over 34% of all documents.

Subtypes

Subtype	Qualified Type	Description
`news`	`article.news`	News reporting and journalism
`blog`	`article.blog`	Blog posts and personal essays
`opinion`	`article.opinion`	Opinion pieces and editorials
`review`	`article.review`	Product, service, or media reviews
`analysis`	`article.analysis`	In-depth analytical content
`press_release`	`article.press_release`	Official press releases and announcements

type_data fields

Field	Type	Required	Description
`authors`	`array<string>`	Yes	Author names
`byline`	`string`	No	Full byline as displayed
`publish_date`	`string` (ISO 8601 date)	Yes	Publication date
`section`	`string`	No	Publication section (e.g., “Technology”, “Politics”)
`word_count`	`integer`	No	Total word count of the article
`reading_time`	`integer`	No	Estimated reading time in minutes
`sources`	`array<string>`	No	Sources cited in the article
`corrections`	`array<object>`	No	Published corrections or updates

Example type_data

{
"type_data": {
  "authors": ["Maria Schmidt", "James Chen"],
  "byline": "Maria Schmidt and James Chen, Technology Desk",
  "publish_date": "2025-03-15",
  "section": "Technology",
  "word_count": 2847,
  "reading_time": 12,
  "sources": [
    "EU Parliament records",
    "Press conference transcript",
    "Industry analyst interviews"
  ],
  "corrections": [
    {
      "date": "2025-03-16",
      "description": "Corrected the implementation timeline from 12 to 18 months"
    }
  ]
}
}

Corrections schema

When an article has been corrected after publication, the corrections array records each correction:

Field	Type	Description
`date`	`string` (ISO 8601 date)	Date the correction was made
`description`	`string`	Description of what was corrected

Notes

The authors field should contain individual author names, not organizational bylines. Use byline for the full displayed attribution.
publish_date is the original publication date, not the SDF conversion date. Use provenance.conversion_timestamp for conversion timing.
reading_time is calculated at approximately 238 words per minute.
For reviews that also have commerce aspects (e.g., product reviews with pricing), use aspects: ["commerce.product"] on the document.