W3C home > Mailing lists > Public > semantic-web@w3.org > February 2005

Re: [Image] How to identify images in an RDF context

From: Richard Newman <r.newman@reading.ac.uk>
Date: Wed, 16 Feb 2005 08:38:23 +0000
Message-Id: <4cddd072d95e1ca77aee915194a16c11@reading.ac.uk>
Cc: semantic-web@w3.org
To: Karl Dubost <karl@w3.org>

Karl,
   Replies in-line.

>> If the EXIF contains a precise date, that's possible, but you would 
>> need some way of mapping back to a file somewhere. I made up a scheme 
>> that uses iPhoto's unique IDs, but some URN based on a timestamp 
>> might be better. Giving it a URI would only work if you (or rather, 
>> the user of the software) has a prefix that they control. Tough one, 
>> that. See later answers for more.
>
> timed stamp urn is fine. I think the time in this case is just a way 
> to make unique the reference more than any kind of serial number. As 
> it's not more practical to have a time stamp than anything else?

I was only leery of timestamps because they might not be unique, and 
because dereferencing is more difficult. That's why I would prefer a 
CIFP (timely!) of timestamp and import location, because both of those 
are mostly inverse functional.

> Abstract entity           = RAW image in the camera

Did I say that? Oops. Not really... the "Work" is the 'photo' you took. 
For example, even rendered as a PNG with a new date attached, the 
original photo (an abstract thing) was taken at a fixed moment, and 
depicts some things, etc, though the "Work" concept is based more on 
intent than that.
The closest physical thing to the work is the raw image on the camera's 
CCD at the instant you took the picture, but it's the abstract thing 
we're referring to.
I'd want to go back and re-read the spec with an eye towards photos, 
which are a bit different to other bibliographic works.

> canonical representation  = JPEG version

The work (the "photograph") is expressed in an expression, which is a 
photograph. This is closer to existence, but still isn't quite there. 
It's closer to say that the expression is the photo you took, on your 
camera, at that precise instance -- your photo of the Eiffel Tower, 
while the work is the concept of a photo of the Eiffel Tower. The 
expression, by nature of its creation by a digital camera, has a 
canonical representation -- the one that was written to CompactFlash.
This JPEG is an item of a manifestation; that chunk of binary that is a 
JPEG. This manifestation can be instantiated multiple times; it remains 
on the CF as I import it into iPhoto, and put it on the Web.

This allows us to refer to multiple levels:
- the photo of the Eiffel Tower
  - might be many if I transform the image to make a derivative work.
- the photo my camera took (these two are very similar for photos)
  - but there may be many, if I re-express the work by cropping...
- the manifestation of each photo (the "template" for JPEGs, if you 
will)
  - of which there are many; different formats etc., just like printings 
of a text
- the items, of which there are many
  - multiple copies of each manifestation (Web, disk)

> I have difficult to see why the Abstract entity has to be the bits in 
> the camera, it could be why your eye has seen through the camera, it 
> can be just the notion of the scene happening that you took in 
> picture? But I guess it doesn't really matter.

Yes, that's pretty much it -- it's a very wooly notion.

>> Each of these corresponds with one of FRBR's 4 layers, and has its 
>> own sets of properties.
> The internal subdivision of Group One entities is important as well.  
> FRBR specifies that intellectual or artistic products include the 
> following types of entities:
> 	▪ 	 the work, a distinct intellectual or artistic creation
> 	▪ 	the expression, the intellectual or artistic realization of a work
> 	▪ 	the manifestation, the physical embodiment of an expression of a 
> work
> 	▪ 	the item, a single exemplar of a manifestation.
> So my photographs belong to the Group 1
> I belong to the Group 2

Yes. I was referring to the types of entities -- the 4 layers being 
work, expression, manifestation, and item. These are the distinctions 
that I think are most useful for modelling this kind of domain.

>> I'd highly recommend digging up some stuff on FRBR, it makes this 
>> kind of thing much clearer. Note, though, that it still doesn't make 
>> it easier to identify the original picture, though if you know the 
>> original import location (iPhoto does) you can actually use that as 
>> an IFP to identify the abstract entity, skilfully avoiding the 
>> problem of giving it a URI! E.g.
>
> For me the original picture is not even the RAW file, because I don't 
> keep it, I just have the JPEG file.

It doesn't need to be -- all you need is some way to reference the work 
when it's not an addressable resource (because the JPEG that comes out 
isn't what you want to talk about). For this, some timestamp or IFP is 
needed, for which I suggest the original location (an address on the CF 
card) and EXIF timestamp. It does mean you can only refer to works for 
photographs that have been manifested, but I'm not sure that's an 
onerous restriction.

> It's when I see that, that I know why I like RDF because it's visual 
> ;) and it's why I prefer to see graphs :)))

:)
Stare at N3 for long enough and you see all sorts of things!

>> I've done a bit of work on modelling FRBR in RDFS and OWL, which I 
>> should also get round to finishing. It might be nice to cast it down 
>> to images as an exemplar.
>
> Yes and to give a step by step tutorial of it. That would be really 
> nice.

I will bump it up my to-do list.

Regards,
-Richard
Received on Wednesday, 16 February 2005 08:39:02 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 07:41:44 UTC