IPTC Media Topics and News Taxonomies Explained

IPTC Media Topics is an openly published, multilingual subject taxonomy for news, maintained by the International Press Telecommunications Council (IPTC). It gives each subject a stable code, such as medtop:20000578 for local elections, so that agencies, publishers and their software describe what a story is about in the same way. It has been IPTC's main subject taxonomy since 2010 and is organised under 17 top-level topics.

For a newsroom, a shared taxonomy solves practical problems: content from agencies arrives already classified, archive search and topic pages work across decades, and analytics can be grouped by subject rather than by URL. Most organisations still need terms of their own, and IPTC has guidance on adding them without breaking compatibility.

This guide covers its structure and languages, how it relates to NewsCodes and the older Subject Codes, how newsrooms use and extend it, and how to map an existing tag list onto it.

Key takeaways

• Media Topics is IPTC's subject taxonomy for news, with 17 top-level topics, labels in 13 languages and variants, and a permanent identifier for every concept.
• IPTC describes it as over 1,200 terms; the September 2026 release contains 1,394 concept records, of which 1,059 are current and the rest are retired but never deleted.
• It replaced IPTC Subject Codes, which remain available but are no longer maintained, and most Media Topics link back to a Subject Code and to Wikidata.
• IPTC recommends extending Media Topics in your own scheme, linked to its terms with SKOS mapping relations, rather than adding terms to the official vocabulary.
• Legacy tags map best when each one is classified first (topic, place, person, genre or internal label) and only topics are mapped to Media Topics.

What IPTC Media Topics is

IPTC publishes a family of controlled vocabularies for news content called NewsCodes. Media Topics is the one for subjects. Its scheme note describes it as IPTC's primary subject taxonomy since 2010, focused on classifying text. It was developed from the earlier Subject Codes, kept the same 17 top-level terms, and extended the hierarchy from three levels to five.

The 17 top-level topics

Below these sit increasingly specific terms. Classical music, for example, is medtop:20000022, reached through arts, culture, entertainment and media, then arts and entertainment, music and musical genre. Each concept has a permanent URI on IPTC's vocabulary server, a short QCode form, a preferred label, a definition, and dates for when it was created and last modified.

How many terms are there?

IPTC's Media Topics page describes a taxonomy of over 1,200 terms. Counting the machine-readable file on IPTC's vocabulary server for the release dated 30 September 2026 gives a more precise picture:

The counts move with each release, so check the live file before quoting a figure. The vocabulary can be browsed as a tree or downloaded as a NewsML-G2 Knowledge Item, RDF/XML, Turtle or Excel. IPTC publishes Media Topics openly on its website and on its vocabulary server at cv.iptc.org.

Languages

IPTC lists Media Topics in 13 languages and language variants: Arabic, British English, US English, Chinese, Danish, French, German, Norwegian Bokmål, Norwegian Nynorsk, Portuguese, Brazilian Portuguese, Spanish and Swedish.

IPTC describes NewsCodes as language agnostic: the code carries the meaning, and only labels and definitions need translating. A German article and an Arabic article tagged with the same code are about the same subject, whatever the labels say. This is one of the strongest arguments for a shared taxonomy in a multilingual archive (see our guide to multilingual archives and AI).

Translation coverage is not uniform. In the September 2026 file, every current concept has British English labels, while some languages have labels for around 980 of the 1,059. If your interface depends on one language, check coverage for that language before launch.

Media Topics, NewsCodes and Subject Codes

NewsCodes is the umbrella name for all of IPTC's controlled vocabularies. They are grouped by purpose (descriptive, administrative, transmission and others) and cover things such as genre, content production roles, video codecs and the digital source of an image. Media Topics is one NewsCodes vocabulary. Genre is another, which is a reminder that "obituary" or "analysis" describes the form of a story and belongs in a separate field from its subject.

Subject Codes was IPTC's original subject taxonomy: about 1,400 terms (more than 1,000 not counting sports competitions) in a hierarchy limited to three levels. IPTC says it is still available but no longer maintained, and recommends Media Topics instead. When its successor's requirements were set in 2007, the goals included allowing a fourth level or deeper and keeping the list to at most 800 terms, excluding sports competitions, so journalists could scan it quickly. It has grown since, but the aim of a list people can actually use remains a good test for any house taxonomy.

Most Media Topics concepts carry a mapping back to a Subject Code, and IPTC provides mappings so that organisations with older tagged content can convert it.

In NewsML-G2, IPTC's news exchange format, a subject is expressed with a QCode:

<subject type="cpnat:abstract" qcode="medtop:20000578" />

The prefix medtop is an alias that a catalogue resolves to the scheme URI http://cv.iptc.org/newscodes/mediatopic/. IPTC's guidelines advise software not to hard-code the alias but to resolve it through the catalogue, because a provider may use a different one.

How newsrooms use Media Topics

IPTC names several users. NewsCodes vocabularies are used by Agence France-Presse, Associated Press and Reuters in their NewsML-G2 services. The Swedish news agency TT and the Norwegian agency NTB use Media Topics to share content with customers, and it is the basis of a content sharing system in Sweden. Ritzau in Denmark and ABC Australia use it as the core of their own vocabularies, extended for their needs. The Austria Presse Agentur uses Media Topics to categorise content in its Basisdienst service.

Inside a newsroom, the same codes typically drive:

• Incoming feeds. Agency copy classified with Media Topics can be routed to desks and filtered without re-tagging.
• Archive search and topic pages. Stable codes let a topic page gather coverage written under different headline styles and tag lists over many years.
• Related content. Shared subject codes are a simple, explainable signal for suggesting related stories.
• Analytics. Performance grouped by topic answers editorial questions that page-level data cannot.
• Automatic classification. IPTC lists several tools that automatically classify content against Media Topics, while warning that some only support the older Subject Codes. Ask any supplier which vocabulary and release it uses.

Our guide to AI in the newsroom covers tagging, related stories and archive search in more detail.

Extending Media Topics with house terms

A global taxonomy cannot hold every local sport, festival or institution. IPTC's NewsCodes Guidelines say exactly this: the working group often hears from organisations that want sports played in only one country or local festivals and holidays. Its example is a publisher specialising in animal stories, which would struggle because Media Topics has a single term for animal (medtop:20000500).

The Q3 2026 update, announced on 16 October 2026, shows why extension matters. It completed a review of the sports branch: 121 definitions rewritten, 17 labels changed, 27 concepts retired, 6 concepts moved and one new term, para sport, added. Hurling was among the retired concepts, with no replacement listed. An Irish publisher that covers hurling every week now needs a house term.

IPTC's recommended approach:

1. Create your own scheme. Do not put new terms in the medtop: namespace, which is reserved for terms maintained by the IPTC working group.
2. Link each house term to Media Topics with a SKOS mapping relation, usually skos:broadMatch to the nearest broader Media Topic.
3. Use skos:exactMatch only for true equivalents. IPTC advises closeMatch or relatedMatch where the meaning is similar but not identical.
4. Propose terms with wider relevance. A term that matters in several countries can be submitted for the official vocabulary.

A house term for hurling might look like this in SKOS Turtle:

@prefix skos: <http://www.w3.org/2004/02/skos/core#> .
@prefix medtop: <http://cv.iptc.org/newscodes/mediatopic/> .
@prefix house: <https://example.com/vocab/sport/> .

house:hurling a skos:Concept ;
    skos:inScheme <https://example.com/vocab/sport/> ;
    skos:prefLabel "hurling"@en-GB , "iománaíocht"@ga ;
    skos:broadMatch medtop:20000822 ;
    skos:changeNote "Created after medtop:20000964 was retired in 2026."@en-GB .

Here medtop:20000822 is competitive sport, so any system that understands only Media Topics still sees the story as sport.

Facets beyond topic

Subject is only one dimension. A practical pattern is to keep other dimensions in separate facets alongside Media Topics rather than forcing everything into one tree:

User needs are a good example of a house facet. The User Needs Model 2.0, developed by a former BBC digital development editor with an analytics company, groups eight needs into four categories: know, understand, feel and do. It describes why someone reads a story, not what it is about, so it belongs in its own field. See our guide to content analytics and user needs.

Mapping legacy tags to Media Topics

Most archives arrive with years of free-text tags. A practical mapping process:

1. Export tags with usage counts and sample headlines. A tag's meaning is clearer from five headlines than from its label.
2. Work in order of usage. Map the tags that cover the most content first.
3. Classify every tag before mapping it. Is it a subject, a place, a person or organisation, a genre, or an internal label such as "homepage pick"? Only subjects go to Media Topics; the rest go to their own facets or are dropped.
4. Choose the relation. Exact, broader (the house tag is narrower than the Media Topic), one-to-many (the tag mixes several subjects), or no match.
5. Re-tag rather than map where tags are ambiguous. A one-to-many tag cannot be converted automatically; each article needs reclassifying, which is where AI-assisted tagging with human review saves time.
6. Record the release you mapped against. IPTC releases Media Topics updates every two to three months, never reuses identifiers and does not change the meaning of existing terms, so a mapping stays valid but should be reviewed after each release for new and retired terms.

For the wider design questions, such as how many facets to have and who approves new terms, see our step-by-step guide to building a content taxonomy.

How Druid Learning helps

Druid Learning is an AI archival processing platform, based in Dublin and New York, that turns content archives into AI-ready datasets. Its Standardization Tool provides real-time content tagging and labelling, so archive and new content can be described consistently. For NPR, Druid processed more than 2 million assets and applied 20+ labels across 8 dimensions, including user needs, article type, location, topic and community. Journalists get suggested tags and related stories, the archive can be searched by label, and audience analytics can be broken down by label in Chartbeat and Google Analytics.

Frequently asked questions

How many terms does IPTC Media Topics have?

IPTC describes Media Topics as a taxonomy of over 1,200 terms under 17 top-level topics. The machine-readable release dated 30 September 2026 contains 1,394 concept records, of which 1,059 are current and 335 are retired. Retired terms are never deleted, so older content tagged with them can still be interpreted. Because IPTC updates the vocabulary every two to three months, check the live file on IPTC's vocabulary server before quoting a figure.

Where can I find and download IPTC Media Topics?

IPTC publishes Media Topics openly on its website and on its controlled vocabulary server at cv.iptc.org. The vocabulary can be browsed online as a tree, or downloaded as a NewsML-G2 Knowledge Item, RDF/XML, RDF/Turtle or an Excel file. Each concept has a permanent URI, labels in up to 13 languages and variants, a definition and creation and modification dates, which makes it straightforward to load into a tagging system.

What is the difference between IPTC Subject Codes and Media Topics?

Subject Codes was IPTC's original subject taxonomy, with about 1,400 terms in up to three levels. Media Topics, released in 2010, kept the same 17 top-level terms but revised the lower levels and allowed deeper hierarchies. IPTC still publishes Subject Codes but no longer maintains them, and recommends Media Topics instead. Most Media Topics link back to a Subject Code, so older tagged content can be converted.

Can I add my own terms to IPTC Media Topics?

Not to the official vocabulary itself, but IPTC expects organisations to extend it. Its guidelines recommend creating your own scheme for house terms, such as local sports or festivals, and linking each one to the nearest Media Topic with a SKOS relation such as broadMatch. Terms that would be useful in several countries can be proposed to IPTC's NewsCodes Working Group for inclusion in the official vocabulary.

How often does IPTC Media Topics change?

IPTC's NewsCodes Guidelines say Media Topics updates are generally released every two to three months. Updates do not reuse identifiers or change the meaning of existing terms; outdated terms are marked as retired, usually with a note on what to use instead. The Q3 2026 update, for example, rewrote 121 definitions, retired 27 sport concepts and added para sport.

Sources

• Media Topics — IPTC
• Media Topic NewsCodes vocabulary — IPTC Controlled Vocabulary Server
• NewsCodes Guidelines — IPTC
• NewsCodes — IPTC
• IPTC NewsCodes 2026-Q3 update released — IPTC
• NewsML-G2 Guidelines — IPTC
• Who's using NewsML-G2 — IPTC
• How to get started with user needs — journalism.co.uk