Hi Andy,
Ed and I have both had read through your discussion paper, and here are the
points we collectively came up with:
1) You say "We do not need to try and answer the question 'What is a
learning object?' here", however we think that many of the points you
raise are based assumptions about what a learning object is, and it would
be as well to spell these out. Specifically, it seems to us that you are
considering a learning object to be online content which can be moved from
one environment to another. We're not clear how your requirements would
relate to books, CD-ROMs or resources like the H.M. Treasury website
http://www.hm-treasury.gov.uk/
We're not objecting limiting the scope of the discussion like
this--discussion of identifiers within this narrower definition is
perfectly valid, and very useful. However, many people will want to create
metadata instances for the type of learning resource listed above (books,
CD-ROMs, websites) so they will need to know that the issues discussed here
perhaps aren't relevant to them.
2) "Why do we need identifiers?" is only a part of the rationale behind
needing this discussion. There is no shortage of identifier schemes: every
repository has one. The questions which (for us) drives this discussion are
"Why should we share identifiers?" and "Do we want to minimise the number
of schemes?". In other words, is anything wrong with the approach possible
with the LOM of every publisher creating their own scheme [the scheme needs
a globally unique identifier--less of a problem than each object needing a
globally unique identifier].
3) We've found your use of the word creator in the issues list a little
confusing. Your use works for institution as a coporate author but we think
that the issues you outline arise from them acting as the initial publisher
(ie the first entity to make the resource available) rather than their role
in creating the resource. We also think that the first of your issues for a
Learning object respository administrator (Lora) is an issue for the
initial publisher, while others are issues for Lora when responsible for
distribution/retail/"secondary publishing" activities.
We also think that there will be an important issue around allocation of
IDs to existing (legacy?) objects, which will concern publishers/Lora.
4) Another issues for resolver system administrators is: what info needs to
be stored and what info needs to be returned to the user? ID and location
are the bare minimum for a resolver to work. Status of the object might be
useful (ie is it still available).
The following relate to the Learning object identifier requirements section:
5) I'm interested to know if these are radically different to the
requirements for digital objects other than LOs.
6) How important is it that identifiers per se should be usable in Web
browsers? Isn't the requirement that resolver services should be usable
through web browsers? The DOI 10.1000/186 itself is not actionable in a web
browser, but doi:10.1000/186 is actionable with the CNRI Handle System
Resolver plug-in (as you say, not an ideal situation--presumably it can be
made actionable through http if a link to a suitable resolver is used).
7) Could you explain your thinking behind requirement 9, identifers should
be URI compliant? Do you mean they should be in the form of URIs or that it
should be possible to include them in URIs (ie should not contain awkward
characters like spaces, ampersands etc).
8) In requirement 7, "they should be assignable in devolved environments",
your reasoning here is clear (we think) but do you really mean "without
reference to a central service"? Don't PURL, DOI, URL all rely on central
services at some point and so anything based on these would? Do you mean
that a publisher should not have to go to a central service repeatedly for
each and every new identifier (but they might have to go to a central
service for a scheme identifier, for an identifier pre-/post-fix, or for
blocks of identifers)?
Hope this helps, Phil
Andy Powell wrote:
> I've put together a short discussion paper concerning identifiers for
> learning objects:
>
> http://www.ukoln.ac.uk/distributed-systems/lo-identifiers/
>
> The intention is to try and build a list of our (the UK HE and FE
> community's) requirements for identifiers, which we can then use to assess
> the suitability of the available identifier technologies.
>
> The paper suggests an initial list of 10 requirements.
>
> Comments welcome...
>
> Andy
> --
> Distributed Systems, UKOLN, University of Bath, Bath, BA2 7AY, UK
> http://www.ukoln.ac.uk/ukoln/staff/a.powell +44 1225 383933
> Resource Discovery Network http://www.rdn.ac.uk/
>
>
--
Phil Barker Learning Technology Adviser
ICBL, School of Mathematical and Computer Sciences
Mountbatten Building, Heriot-Watt University,
Edinburgh, EH14 4AS
Tel: 0131 451 3278 Fax: 0131 451 3327
Web: http://www.icbl.hw.ac.uk/~philb/
|