Is there a namespace to describe mimetypes and encodings?

Question

What I would like to be able to say is:

<http://totl.net/> dct:format <http://example.org/mimetype/text/html> .
<http://totl.net/> dct:format <http://example.org/encoding/utf-8> .

Of course I could just invent my own namespace, but I'd rather find the "right" one.

If there is no right one, there should be.

Answer 1

Hi,

for mime types, check out http://mediatypes.appspot.com/.

Of course even that is not as "official" as it could be. IANA hasn't (AFAIK) yet set anything up for providing URI:s and RDF for their registries, which I think they should. In order for that to happen, we as a community have to build a public demand, and/or advocate to them diligently [*].

Otherwise, it's quite common to give the media type as a literal or a bnode using DC. Something like (a bnode version):

<rdf:Description xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
                 xmlns:dct="http://purl.org/dc/terms/"
                 xmlns:dctype="http://purl.org/dc/dcmitype/"
                 rdf:about="http://example.org/document">
  <dct:hasFormat>
    <dctype:Text rdf:about="http://example.org/document.html">
      <dct:format>
        <dct:IMT>
          <rdf:value>text/html</rdf:value>
        </dct:IMT>
      </dct:format>
    </dctype:Text>
  </dct:hasFormat>
</rdf:Description>

([*] See e.g. the discussion at http://lists.w3.org/Archives/Public/public-lod/2010Apr/0056.html for a related topic concerning IANA link relations.)

Answer 2

0

The NEPOMUK Information Element Ontology has a :mimeType predicate. Is has a xsd:string as range; the domain might be a problem, as it is defined as a :InformatationElement which may not be compatible with your application.

link

answered 14 Mar '10, 10:31

Egon Willigh...
1.5k●5●12
accept rate: 12%

1

Thanks. I'm utterly surprised that there isn't a URI scheme for it... Except that you probably can't use "/" in an N3 or XML element, which leaves me with the option of changing "/" to "_" or using full URIs (so good to keep it a short namespace.)

Perhaps http://ns.eprints.org/mime/text/plain and http://ns.eprints.org/enc/utf-8

It seems obvious to me that these things require URIs not strings as them being equal means they are the same concept, not true of strings.

(15 Mar '10, 10:07) Christopher ...

Answer 3

0

Hi,

I don't know about the encoding but for the file format and the mime type, you can look here: http://p2-registry.ecs.soton.ac.uk/ there might be something interesting. The scheme does something like http://p2-registry.ecs.soton.ac.uk/pronom/Identifier/text/html

link

answered 18 Apr '10, 12:13

florent
79●2
accept rate: 0%

Answer 4

You shouldn't feel intimidated to create your own predicates in case you find that other people's predicates aren't perfect for your application.

You can always use some kind of inferencing (such as rdf:subPropertyOf) to map your crisp private predicate to 'vernacular' predicates that other people use. The big picture is that "anyone can say anything about any subject", and you shouldn't let a lack of the right predicate slow you down.

Answer 5

My approach would be a hybrid of the dct:IMT approach and the http://purl.org/NET/mediatypes/ namespace:

<http://example.com/page.html> dcterms:format <http://purl.org/NET/mediatypes/text/html> .

<http://purl.org/NET/mediatypes/text/html> a dcterms:IMT ;
  rdf:value "text/html" ;
  rdfs:label "HTML document" .

The reason is that the http://purl.org/NET/mediatypes/* sadly does not redirect to an RDF representation, and also the RDF at dump.rdf is very sparse. With my suggestion here you provide as much as you know about the format directly, but by using the PURLs you avoid all those bnodes and can do some simpler SPARQLs.

Is there a namespace to describe mimetypes and encodings?

Follow this question

Upcoming Events

Related questions