Metadata

DFG Digitisation Guidelines

The "DFG Practical Guidelines on Digitisation" (​German, ​English) discuss some metadata issues.

Dublin Core

Dublin Core Metadata Initiative (links with a specified date point to a specific document verson, links without date point to the most recent version):

• ​homepage, ​overview
• DCMI Metadata Terms: ​DCTERMS (main document), ​REVISIONS, ​DOMAINS
• DCMI Namespace Policy: ​NAMESPACE
• DCMI Abstract Model: ​DCAM
• DC and RDF: ​DC-RDF, ​DC-RDF-NOTES
• DC and XML: ​2003 XML Guidelines, may soon be superseded by the proposed recommendation ​2008 XML Guidelines, ​2008 additional notes (there is also a ​2006 working draft)
• ​software (mainly online tools), ​FAQ

Some findings: Within the original 15 elements, creator now refines contributor, and source refines relation (i.e. searching for a contributor should also find a creator, etc.); one can still use dc: for the original 15 elements, but is encouraged to use dcterms: instead. dcterms:date has range Literal, dcterms:language has range LinguisticSystem (​DCTERMS). DateScheme and LanguageScheme were never officially declared and are no longer used; ISO-639-3, W3CDTF (​REVISIONS).

Metadata in our Schema

The ​metadata module in our Relax NG compact schema.

Some questions:

1. The definitions

   dc.language = element dcterms:language { element rdf:Description { element dcq:languageScheme { "ISO 639-2" }, rdf.value } }
   dc.date = element dcterms:date { element rdf:Description { element dcq:dateScheme { "ISO 8601" }, rdf.value } }
   

seem to be obsolete. Especially dcq seems obsolete (see ​NAMESPACE). I have replaced them by something which I believe is in accordance with Recommendation 7 in the ​2003 XML Guidelines:

dc.language = element dcterms:language { attribute xsi:type { "dcterms:ISO639-3" }, text }
dc.date = element dcterms:date { attribute xsi:type { "dcterms:W3CDTF" }, text } 

(Maybe I should have left <rdf:value> instead of "text"? But then, consequently ALL "text" in dcterms definitions should in fact be <rdf:value>.)

However, information is sparse. The syntax encoding schemes dcterms:W3CDTF and dcterms:ISO639-3 do exist according to ​DCTERMS, but an up-to-date opinion about how to use syntax encoding schemes is difficult to obtain. The proposed recommendation ​2008 XML Guidelines uses dcterm:date only like this:

<dcterms:date>2005-05-05</dcterms:date>
<dcterms:date rdf:datatype="http://www.w3.org/2001/XMLSchema#date">2005-05-05</dcterms:date>

Any suggestions?

2. Should we rename our <metadata> element? In all examples it is <ref:Description>. However, in most examples it seems to mean the repository version of the metadata, not the version in the xml file itself.

3. Do we need something like <foaf:Person> for creators, contributors, etc.? (I guess not; it doesn't seem have any particular advantage.)