Type Taxonomy

Overview

SDF classifies every document using a two-level hierarchy: parent_type.subtype. There are 10 parent types and 50+ subtypes covering the full range of web content.

In production deployment across 2,335 documents, 74 unique type combinations were observed. The type normalization cascade corrected 63 non-canonical type inventions to achieve 100% taxonomy conformance.

Taxonomy tree

article
├── news
├── blog
├── opinion
├── review
├── analysis
└── press_release

documentation
├── tutorial
├── api_reference
├── guide
├── faq
├── changelog
└── troubleshooting

commerce
├── product
├── category
├── comparison
├── marketplace
├── service
└── pricing

discussion
├── forum
├── qa
├── comment_thread
├── review_thread
└── poll

reference
├── encyclopedia
├── dictionary
├── legal
├── academic
├── specification
└── wiki

data
├── dataset
├── statistics
├── report
├── dashboard
├── financial
└── scientific

code
├── repository
├── package
├── snippet
├── gist
├── notebook
└── documentation

profile
├── person
├── organization
├── place
├── product_profile
└── portfolio

event
├── conference
├── meetup
├── webinar
├── concert
├── sports
└── workshop

media
├── video
├── podcast
├── music
├── image_gallery
├── livestream
└── animation

Reference table

Parent Type	Subtypes	Description	type_data Reference
`article`	news, blog, opinion, review, analysis, press_release	Authored content for publication	article
`documentation`	tutorial, api_reference, guide, faq, changelog, troubleshooting	Technical / instructional content	documentation
`commerce`	product, category, comparison, marketplace, service, pricing	Product and service listings	commerce
`discussion`	forum, qa, comment_thread, review_thread, poll	Community Q&A and forums	discussion
`reference`	encyclopedia, dictionary, legal, academic, specification, wiki	Encyclopedic and reference material	reference
`data`	dataset, statistics, report, dashboard, financial, scientific	Data-centric content	data
`code`	repository, package, snippet, gist, notebook, documentation	Software and code content	code
`profile`	person, organization, place, product_profile, portfolio	Entity profiles	profile
`event`	conference, meetup, webinar, concert, sports, workshop	Event listings and schedules	event
`media`	video, podcast, music, image_gallery, livestream, animation	Audio/video/multimedia	media

Production statistics

From production deployment over 2,335 documents:

Metric	Value
Total documents processed	2,335
Unique parent types observed	10 (all)
Unique type combinations observed	74
Non-canonical types corrected	63
Taxonomy conformance after normalization	100%

The most frequently observed types in production:

Type	Count	Percentage
`article.news`	487	20.9%
`article.blog`	312	13.4%
`documentation.guide`	198	8.5%
`commerce.product`	176	7.5%
`reference.wiki`	154	6.6%
`code.repository`	143	6.1%
`discussion.qa`	128	5.5%
Other types	737	31.5%

Type System Spec

See Type System for the normalization cascade and aspect system.

Document Model

See Document Model for how type_data fits into the document structure.