And if several people make references to data in D1 that are no longer
valid in D2? It may be clear that a change has occurred at the
repository end, but it is left to the reader to carefully check dates,
and even if they realise the change has occurred, it is not discoverable
whether or not the reference is still valid because the D1 document is
gone and can't be compared. (Sometimes people do refer to early versions
and should be free to do so: who are we to decide that it is an
unreliable business?) It is also not easy to discover by machine, as it
would be if repositories kept all versions in a time-stamped manner and
the URIs reflected this. Your climategate example is a case in point, I
think. If authors really want to stop earlier versions being found by
the public, that doesn't mean they should be deleted, as the
administrator should still have an audit trail: for example, where data
has been altered unethically.
How does freedom of information law impact upon restricted items, by the
way?
Talat
Leslie Carr wrote:
> It's worth remembering that some repositories represent published work only, and so the need for revisions will be fairly low and the policies will be fairly tight.
>
> Other repositories hold work in progress (active research) and so have a much more fluid policy on changes to metadata and the replacement/deletion or addition of data.
>
> For example, for a conference paper, an ECS author may first create a record (R) and deposit a document that is the version of the article that was submitted for reviewing (D1). Later (when the paper is accepted) they might remove the original paper and add the final accepted version (D2) to the same record. The record will then still have a single document (now D2), but it will be clear that it is a different document to the original (D1 no longer exists). After the conference takes place, they may then subsequently add the slides for the presentation as a another document (D3) attached to the same record.
>
> Our repository is careful to store all the changes to the metadata in an audit trail. However, it is up to the author to decide whether or not to keep the original documents/files. If they delete them, they are gone.Although this may seem irresponsible from a library POV, I think it is the only approach in a near-to-research repository.
>
> Of course, climategate may force me to change this opinion.
> --
> Les
>
> On 18 Feb 2010, at 09:44, Chris Rusbridge wrote:
>
>
>> This has been an interesting thread on the relatively small need for changes to versions in repositories (see James Toon's 5% figure). It prompts me to wonder, and to ask, what repository policies are towards changes represented by:
>>
>> a) errata/corrections for minor errors
>>
>> b) legal issues (including but not limited to copyright)
>>
>> c) revised editions (not at all common for scholarly articles, but common for some other kinds of works)?
>>
>> Anything explicit? I would imagine the answer at (a) should be to add a note on the erratum and a new version, retaining the original marked as deprecated; for (b) either a similar response (but probably with the original suppressed) or complete withdrawal; and (c) probably a new record linked to the old (I don't think most repository metadata is up to describing two editions with perhaps different editors, dates, ISBNs etc, but I could be wrong!).
>>
>> No doubt SHERPA has dealt with this already...
>>
>> --
>> Chris Rusbridge
>> Director, Digital Curation Centre
>> Email: [log in to unmask] Phone 0131 6513823
>> University of Edinburgh
>> Appleton Tower, Crichton St, Edinburgh EH8 9LE
>>
>> The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.
>>
>
>
--
Dr Talat Chaudhri
------------------------------------------------------------
Research Officer
UKOLN, University of Bath, Bath BA2 7AY, Great Britain
Telephone: +44 (0)1225 385105 Fax: +44 (0)1225 386838
E-mail: [log in to unmask] Skype: talat.chaudhri
Web: http://www.ukoln.ac.uk/ukoln/staff/t.chaudhri/
------------------------------------------------------------
|