Skip to content
Property · DCMI Metadata Terms: properties

dct:description explained: abstracts, summaries and contents lists

Published Formatory editorial team
http://purl.org/dc/terms/description

An account of the resource.

Description may include but is not limited to: an abstract, a table of contents, a graphical representation, or a free-text account of the resource.

The dct: prefix that shows up in DCAT dataset catalogs, in Schema crosswalks and in countless Turtle files stands for the DCMI Metadata Terms namespace, http://purl.org/dc/terms/. Library documentation tends to write the same namespace as dcterms:. Whichever prefix a file declares, dct:description and dcterms:description are one property with one URI, and its definition is about as open as a metadata definition can be: an account of the resource.

What an account can be

The usage comment lists the expected forms: an abstract, a table of contents, a graphical representation or a free-text account. In practice the free text dominates. A cataloger’s note about a photograph, a data steward’s paragraph on how a dataset was collected and the publisher’s blurb for a book all fit.

The graphical option is the one people forget. A thumbnail of a map or a cover image is a description in the DCMI sense, usually given as a URI to the image. Most systems handle thumbnails through their own fields, but the option explains why the property has no literal range: unlike dct:title or dct:identifier, its value may be a resource.

Abstracts and tables of contents

Two sub-properties narrow the meaning. dcterms:abstract is a summary of the resource, and dcterms:tableOfContents is a list of its subunits. Both refine dcterms:description, which in turn refines the original dc:description.

The split earns its place in repositories where several kinds of description sit side by side. A thesis record might carry the author’s abstract, a chapter list and a short note added by the library. Typing them separately lets an interface show the abstract prominently, collapse the contents list and keep the note for staff. When the record is exported to simple Dublin Core, all three become dc:description values, so nothing is lost for systems that only know the fifteen elements.

Writing descriptions that help

A useful description tells a reader what the title and the other fields cannot. For files and datasets that tends to mean:

  • the scope and coverage (period, place, population, sample size where known)
  • how the content was produced: instrument, method, source collection
  • known gaps, errors or restrictions
  • the structure of a multi-part resource

Common faults run the other way. Descriptions that repeat the title, descriptions stuffed with keywords meant for search engines, and boilerplate pasted onto thousands of records (“This item is part of the digital collections”) all make search results worse. Boilerplate is the most damaging, because it makes every record match every query that contains its words.

Topical terms belong in dct:subject, and the kind of resource in dct:type. A description can mention both in prose, but it should not be the only place that information lives.

Language and markup

RDF literals can carry a language tag, and descriptions benefit from it more than most properties. "..."@en and "..."@fr on the same resource let an interface pick the version matching the user’s language, and they make the language of the metadata explicit, which is not always the language of the resource.

Markup is the last practical question. HTML in a description renders on the site that wrote it and shows up as literal tags almost everywhere else, including OAI-PMH harvesters and DCAT portals. Line breaks survive better, and a URL in plain text is still clickable in most interfaces. Open data portals that harvest DCAT records usually place the description directly under the dataset title, often cut to a few lines, so the first sentence is worth writing as if it might be the only one read.

Example in RDF

@prefix dct: <http://purl.org/dc/terms/> .

<https://example.org/datasets/bridge-sensors>
    dct:title "Strain gauge readings, Elm Street bridge"@en ;
    dct:description "Hourly strain readings from twelve sensors fixed to the main girders, January 2021 to December 2023. Gaps during sensor replacement are flagged in the quality column."@en ;
    dct:abstract "Three years of structural monitoring data from a steel road bridge."@en .

Questions about dcterms:description

What is dct:description?

dct:description is the common prefixed form of dcterms:description, the DCMI Metadata Terms property for an account of the resource. It can hold an abstract, a table of contents, a summary written by a cataloger or a reference to a graphical representation.

Are dct: and dcterms: different vocabularies?

No. Both prefixes point to the same namespace, http://purl.org/dc/terms/. DCAT and many other W3C specifications write dct:, while library and repository documentation more often uses dcterms:. The URIs of the properties are identical.

When should dcterms:abstract be used instead?

dcterms:abstract is a sub-property of dcterms:description for a summary of the resource, typically the author's abstract of an article or thesis. Using it lets systems separate author-supplied abstracts from catalog notes, while still treating both as descriptions.

Can dct:description contain HTML?

Nothing in the definition forbids it, but most consumers treat the value as plain text and display tags literally. Plain text with paragraph breaks is the safer choice, and markup intended for one website rarely survives harvesting into another system.

Should a description repeat the title?

No. A description that restates the title adds nothing for users and wastes the space search results give to it. It should say what the title cannot: scope, coverage, method, notable gaps or the structure of the content.

More from DCMI Metadata Terms: properties

All 13 in this group