John Smith wrote:
> Matthew,
>
> You may be correct that we are still focussed too closely on the
> article as the basis of communication (as in the past we were too
> focussed on the journal as the carrier and quality controller).
> However, I still feel that any new system still needs to play the
> same roles as the old, e.g., quality control, recognition of work
> done, etc. However do you feel that even these (what I see as)
> fundamental requirements are in fact artefacts of the old system that
> do not need to be carried over to the system you envisage?
>
> You did say you wanted to 'up the level of abstraction' :-).
If people are interested, I'm sure we can have a get-together sometime
at OR08, and we can ruminate about this ;-)
I've moved to liking the SWAP model: Research Grant => piece of work =>
version of that work => copy of that version.
eg:
Grant: "Investigate how different engines affect performance in land Rovers"
From this, I produce two articles: "How to fit a V8 in place of a
diesel engine" and "Electric engines Off Road". I also have a
poster-presentation at the "Environmentanlists Conference".
... this is three pieces of work from one grant
The "Electric Engines" article has 3 versions: my submitted draft (ie,
pre peer-review); the "authors final version" (post peer-review); and
the "publishers final copy".
.... this is three instances of the same piece of work
Finally, my "authors final version" has three file copies: my OpenOffice
file, the PDF sent in, and an unformatted ASCII-text version.
In the ideal world, all [3 x 3 x 3 =]27 items will be deposited, and
(most of them?) made available through OA.
My argument is not with this at all... this is fine.
My argument is how these items are deposited: 27 separate deposits is
not the way to do it.
I would contend that one should define one research grant (with its
associated metadata). From that single grant, one links to the three
pieces of work (each with their own specific items of metadata). From
each of these entries, you create three sub-records: the instances. The
instances have very little data specific to themselves (a date,
peer-reviewed/published status', etc [we know the journal the piece of
work was written for already]).
Finally, from each instance we hang three files, along with some
preservative metadata (size, format, MD5 checksum, date deposited, etc)
In *this* system, the researchers interaction with the system is limited
to half-a-dozen fields, all blindingly obvious, and all obviously relevant.
(but until someone can make a system like this, I'll keep my current
repository running, thank you for asking :) )
--
Ian Stuart.
Devloper, the Depot
EDINA,
The University of Edinburgh.
http://edina.ac.uk/
|