Common spec/interface for file metadata

[Jamie McCracken]

Daniel P. Berrange wrote:
> 
> 
> There are multiple levels of Dublin Core compliance. At the basic level,
> the terms are flat
> 
>   contributor
>   coverage
>   creator
>   date
>   description
>   format
>   ...
> 
> they can, however, be further refined
> 
>   description
>     - abstract
>   date
>     - created
>     - copyrighted
>     - submitted

I have looked at it and my first thoughts were they are quite generic 
and vague.

For me the metadata should be as obvious to the user as possible so they 
can enter it in searches and also when they view metadata for a file its 
consistent with the metadata they have already been exposed to. For 
example XMMS exposes metadata for an mp3 as say "Artist" yet Dublin Core 
uses the term "Creator" (it also uses this term for Author of a document 
whilst both Abiword and Evince use "Author". "Creator" in Evince means 
the version of the library used to create the pdf thus further leading 
this term into confusion).

Using the flat Dublin Core wont work because metadata keys need to be 
unique for a file so you cant have two "description" fields even though 
they may relate to different things (we are using hashtables for these)

Also a lot of the metadata does not fit nicely into the 15 generic 
Dublin Core types.

FOAF like RDF is overkill for this too (we really dont need extremely 
hierarchical types that are designed for XML representation here).

Simple metadata names are really all thats required and they should 
match whats currently shown in applications as much as possible. I would 
like to use a standard for naming metadata if possible but it must not 
be too technical or vague for users.


-- 
Mr Jamie McCracken
http://www.advogato.org/person/jamiemcc/