Quoting Lee Passey <[email protected]>:

> So it seems to me that an OL Work record is best exposed as a FRBR
> Expression, and an OL Edition is best exposed as a FRBR Manifestation. I
> suspect that there are some attributes|properties that should be moved in
> the schema from one record to the other for the best match, but OL has no
> real commitment to any specific schema so this could be easily
> accomplished. I think this mapping of OL records to FRBR objects is good
> enough to go with for the time being.


I took another look at the OL Work "type"
   http://openlibrary.org/type/work

I'm not sure if all of the elements there are actually being used, but  
it seems to me to be a good example of theory v. practice. And more  
proof that FRBR is a conceptual model, not a data model.

The things that don't strictly belong in a frbr:Work description that  
are in the OL type:work are:

original language
translated title
first publish date
cover image
first sentence

I'm not sure if translated title and original language are being used  
at this point. Translated title is not = titles of manifestations, but  
is there because there are people who want to actually translate the  
titles for users who do not read the language of the Work. (I admit I  
am somewhat hesitant about this, but some people feel strongly about  
it.)

The first publish date is used in the display of works resulting from  
a search to give users an idea of the epoch represented by the work. I  
believe it is in the ol/work because that's more efficient than  
calculating the earliest date on-the-fly for that particular display.  
Cover image and first sentence are part of the UI, but not FRBR; there  
is an attempt in OL to give users something more than just an author  
and title in the display, something more explanatory, more interesting.

I don't see the ol/work as representing the expressions of the work,  
and at most it could represent the original expression, (a concept  
which I think is oddly missing from the cataloging rules), not all  
expressions. So if a work has been translated into 30 languages, this  
type/work at best could only represent one of them. The others are  
represented in the type/edition

    http://openlibrary.org/type/edition

which is pretty much = frbr:manifestation+expression. It is expression  
because it contains the genre (in this case always book) and language,  
plus any contributing agents (illustrators, translators, etc.), and  
manifestation because it contains the particular publication details.

So the way I read it, the ol/work is frbr:work plus some locally  
needed elements (that are not being exported in the RDF output), and  
the ol/edition is manifestation+expression. The only frbr:items  
represented in the catalog are digitized books, and they are linked  
from the ol/edition but have their own metadata in the related record  
in the Internet Archive.

As I said above, there is the question of theory v. practice, and we  
shouldn't expect the internal data format to strictly follow something  
like FRBR. If only for reasons of efficiency, the internal format may  
need other elements, such as the links from ol/work to ol/editions  
(where FRBR only recognizes links from expressions to works, not vice  
versa).

Do not read this as a support of the FRBR WEMI on my part... I am  
still not at all convinced that the division of entities and the rules  
for their relationships are viable in practice. I do think it is a  
good idea to try to understand what the IFLA FR groups are attempting,  
although I find it to be overly rooted in current cataloging practice.

kc

-- 
Karen Coyle
[email protected] http://kcoyle.net
ph: 1-510-540-7596
m: 1-510-435-8234
skype: kcoylenet

-- 
Karen Coyle
[email protected] http://kcoyle.net
ph: 1-510-540-7596
m: 1-510-435-8234
skype: kcoylenet

_______________________________________________
Ol-tech mailing list
[email protected]
http://mail.archive.org/cgi-bin/mailman/listinfo/ol-tech
To unsubscribe from this mailing list, send email to 
[email protected]

Reply via email to