Skip to content
Class · DCMI classes and encoding schemes

dcterms:FileFormat: the FileFormat class in DCMI Terms, with examples

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

A digital resource format.

Examples include the formats defined by the list of Internet Media Types.

“Fileformat” is one of those words that looks like a typo and is in fact a precise piece of vocabulary. In the DCMI Metadata Terms namespace, dcterms:FileFormat is the class of digital resource formats: PDF, TIFF, WAVE, a particular version of EPUB. It names the kind of thing a format is, so that a statement like “this file is in format X” has a typed object on the right-hand side.

A class for formats, not for files

The most common misreading is to type the file itself as a FileFormat. A scan of a herbarium sheet is a digital object; TIFF 6.0 is the format it is encoded in. The class belongs on the second resource. The scan then links to it with dcterms:format, whose range is dcterms:MediaTypeOrExtent. Because FileFormat sits under MediaType, which sits under MediaTypeOrExtent, a FileFormat is a valid value for that property.

That chain of classes looks bureaucratic, and it is useful. It lets one property carry three different kinds of answer (an extent, a physical medium, a digital format) while still allowing a validator to tell them apart.

Where the format resources come from

DCMI does not publish a list of file formats. The comment on the class points to Internet Media Types as an example of where members can be drawn from, and the scheme dcterms:IMT refers to that IANA registry. A value like image/tiff is a reasonable starting point for discovery systems.

For preservation it is often too coarse. The image/tiff media type does not say whether a file is baseline TIFF, a GeoTIFF or a TIFF wrapping JPEG-compressed strips, and those differences matter when planning a migration. Format registries fill the gap: the National Archives’ PRONOM assigns identifiers such as fmt/353 (TIFF) to formats and their variants, and the Library of Congress Sustainability of Digital Formats site describes formats with stable fdd identifiers. Either can serve as the URI of a FileFormat resource, and tools such as DROID or Siegfried report PRONOM identifiers directly after identification. The file format identification guide walks through that step.

Recording format with enough precision

Three levels tend to appear in a well-made record:

Level Example Typical source
Media type application/pdf web server header, IANA registry
Format version PDF 1.7 identification tool, format registry
Profile PDF/A-2b validator output

Not every level is always available. A collection of born-digital office documents may only have trustworthy media types until a batch identification run is done. What matters is that the value says which level it is, which is easy when each level has its own URI and hard when “PDF” is typed into a text box.

FileFormat next to the older dc:format

The legacy element dc:format has no range, so it takes strings, and most OAI-PMH records fill it with a media type or a word such as “image”. Mapping such records into linked data usually means looking each string up once, building a small table from string to FileFormat URI, and replacing the literal. “JPG”, “jpeg”, “image/jpg” (not a registered media type) and “image/jpeg” will all show up, and all four should land on the same resource. The lookup table itself is worth keeping under version control, because the next harvest will bring new spellings and the mapping should grow with them.

A physical carrier is a different case. A 3.5-inch floppy disk is a dcterms:PhysicalMedium, recorded with dcterms:medium, even when the record also describes the files that were read off it.

Example in RDF

@prefix dcterms: <http://purl.org/dc/terms/> .
@prefix rdfs:    <http://www.w3.org/2000/01/rdf-schema#> .

<https://example.org/formats/tiff-6> a dcterms:FileFormat ;
    rdfs:label "TIFF, revision 6.0" .

<https://example.org/scans/plate-0042.tif>
    dcterms:format <https://example.org/formats/tiff-6> ,
                   <https://www.iana.org/assignments/media-types/image/tiff> .

Questions about dcterms:FileFormat

What is dcterms:FileFormat?

It is a class in the DCMI Metadata Terms namespace whose members are digital resource formats, such as PDF, TIFF or a specific version of a format. A resource typed as dcterms:FileFormat is a format, not a file; files point to it through dcterms:format.

How is FileFormat related to MediaType?

dcterms:FileFormat is declared a subclass of dcterms:MediaType. A media type covers both file formats and physical media, so every file format is a media type, while a physical carrier such as a microfiche is a media type that is not a file format.

Is a MIME type the same as a FileFormat?

A MIME type such as image/png identifies a format at the level the Internet Media Types registry draws, which is often coarse. One media type can cover several format versions, so preservation systems usually record a finer format identifier alongside it.

Can I use a plain string such as 'PDF' as a FileFormat?

With dc:format a string is acceptable. FileFormat is a class, so its members should be resources with URIs. If no format registry URI is available, a local URI with a label is still better than a free-text value for anything machine-processed.

Does FileFormat cover file format versions?

The class does not say how granular a format must be. Whether TIFF as a family or TIFF 6.0 as one version is the right member depends on the registry or vocabulary used. Preservation work generally needs the version level.

More from DCMI classes and encoding schemes

All 10 in this group