W3C home > Mailing lists > Public > www-rdf-interest@w3.org > October 2003

Re: Attaching (embedding?) RDF

From: Martin May <maym@foobar.lu>
Date: Tue, 28 Oct 2003 12:18:34 +0000
Message-Id: <D9F3D650-0940-11D8-A3AD-0030656AB632@foobar.lu>
Cc: <www-rdf-interest@w3.org>, "Rahul Singh" <kingtiny@cs.cmu.edu>
To: "Jeremy Carroll" <jjc@hplb.hpl.hp.com>


That sounds like a good solution for avoiding custom handlers with 
option 3. The main problems I see with it are:

- it would not always work (imagine a document that describes RDF, it 
will have matches that are NOT the metadata of the document).
- you're assuming that metadata embedded in binary files will be 
textual (it might be encoded, in which case the search won't match).

I've thought about this for a while, and option 3 seems less and less 
attractive to me, as there won't be a generic way of handling the 
metadata, which is what I want.

Does anybody have thoughts on option 4, consisting of creating some 
kind of bundle (e.g. a zip or jar file) that binds the file and it's 
rdf metadata file together?

Thanks,

Martin


On 28 Oct 2003, at 09:57, Jeremy Carroll wrote:

>
> A workable hack on the embedding front (option 3) is to do a linear 
> search
> of the file for the rdf:RDF tag, ideally this needs to be smart i.e. a
> search for
> "<" + NCName + ":RDF" + whitespace + other xml attributes + "xmlns:" +
> NCName + whitespace + "=" + whitespace + ["'] +
> http://www.w3.org/rdfnamespaceuri# +["'] + other xml attributes + ">"
>
> One needs to assume UTF-8 or UTF-16 encoding.
>
> A linear search for the namespace uri can be quite efficient (I don't 
> have
> the reference for the technique for searching for a long fixed 
> substring,
> but you only have to look at a few of the characters). You then need 
> to work
> back from there to see if the rdf:RDF matches.
>
> The idea is that at first approximation all embeddings are the same 
> and one
> doesn't need to know the master file type.
>
> Jeremy
>
>> -----Original Message-----
>> From: www-rdf-interest-request@w3.org
>> [mailto:www-rdf-interest-request@w3.org]On Behalf Of Martin May
>> Sent: 27 October 2003 20:14
>> To: Rahul Singh
>> Cc: www-rdf-interest@w3.org
>> Subject: Re: Attaching (embedding?) RDF
>>
>>
>>
>>
>> Yes,
>>
>> I actually looked at XMP, and it seems that they are using method #3
>> (please correct me if I'm wrong). IMHO the drawback of not being able
>> to store metadata in a generic way and having to write custom adaptors
>> for every file type is quite a big one. On top of that, some file 
>> types
>> don't support embedded metadata at all.
>>
>> I was hoping that somebody would know of some alternatives to the ones
>> that I presented or know of methods to overcome the drawbacks.
>>
>> Thanks,
>>
>> Martin
>>
>> On 27 Oct 2003, at 18:43, Rahul Singh wrote:
>>
>>> Have you looked at Adobe XMP?
>>>
>>> http://www.adobe.com/products/xmp/main.html
>>>
>>> Cheers,
>>>
>>> Rahul
>>>
>>> Martin May wrote:
>>>
>>>>
>>>>
>>>> I would like to attach RDF metadata to files. So far I have
>>>> identified the following options for doing so, all of which have
>>>> drawbacks:
>>>>
>>>> 1. given a file foo.pdf, store the metadata in a file foo.pdf.rdf in
>>>> the same directory
>>>>
>>>> Main drawbacks:
>>>>
>>>> - if the file is moved or renamed, the metadata must be moved with 
>>>> it
>>>>
>>>> 2. store the metadata in a central repository, with a link (URI) to
>>>> the file
>>>>
>>>> Main drawbacks:
>>>>
>>>> - again, if the file is moved or renamed, the link becomes invalid
>>>> - requires a central repository
>>>>
>>>> 3. embed the metadata in the file format
>>>>
>>>> Main drawbacks:
>>>>
>>>> - doesn't work with every file type
>>>> - different handling for each file type required
>>>>
>>>> 4. create an archive file which groups the file and its metadata
>>>> (e.g. jar)
>>>>
>>>> Main drawbacks:
>>>>
>>>> - requires unpackaging and repackaging every time the file needs to
>>>> be accessed
>>>>
>>>> Does anybody have any insights or ideas on how to improve the
>>>> suggested methods, or provide me with other alternatives?
>>>>
>>>> Best regards,
>>>>
>>>> Martin May
>>>> University of Aberdeen
>>>>
>>>>
>>>
>>
>>
>
Received on Tuesday, 28 October 2003 08:37:57 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:52:02 GMT