Leslie Carr wrote:
> On 13 Feb 2008, at 11:31, John Smith wrote:
>> As I understand IRs they are meant to be true repositories of research
>> outputs. Any version of full text entered should be retained and
>> if/when a newer version of the full text is entered it should be as a
>> new item with a new item number. However, this leads to duplication of
>> titles (if not content) and could cause a problem when the IR is used
>> as the source for other services like staff publication lists and
>> annual publication lists, etc.
>>
>> How many UK IRs keep to the strict repository rule and retain all
>> versions? If you don't do you edit/update existing records or
>> duplicate and delete older records when new versions are added?
>
> We've used the EPrints "clone record" facility to create version chains
> of eprints. See http://eprints.ecs.soton.ac.uk/8669/ as an example -
> there is an alert warning that the currently displayed eprint isn't the
> most up-to-date version and if you scroll down a little you will see
> links to all the available versions.
This display is quite similar to what we do at arXiv in that we display a
warning when an older version is accessed, e.g.:
http://arxiv.org/abs/cs/0602009v1
which shows the "warning" by bolding/coloring the latest version part of
the line:
(Submitted on 6 Feb 2006 (v1), last revised 13 Feb 2008 (this version, v77))
nothing is bolded for access to the "current" version (via URL without version
number) or explicit access to last version:
http://arxiv.org/abs/cs/0602009
http://arxiv.org/abs/cs/0602009v77 (current as of 2008-02-20)
I think this article wins arXiv's "most revised" award (not usually a good
sign in terms of quality assessment). Most papers have just one or a couple of
versions.
> This isn't universally used by everyone, and we don't ask people to
> deposit multiple versions any more for several reasons:
> (a) the versioning facility gets abused by people who want to falsely
> duplicate an eprint just to get the author list copied for a new,
> unrelated publication.
> (b) people get confused between versions of publications and versions of
> records
> (c) people get confused between versions of publications and new
> publications
We usually insist that submitters use our replacement facility to create new
versions, rather than making new submissions, if the revision is within a
couple of years of original submission. If later than that we usually let the
submitter decide.
In the mechanics of the submission process, metadata is taken from the
previous version to populate the upload form which sounds somewhat like the
'clone' Les describes. This can't be abused for new submissions as we have a
different processes.
Cheers,
Simeon
> --
> Les
>
|