> > OAI-ORE. I agree with the need to describe aggregations in some form
> > vs. individual objects and OAI-ORE is worth considering. But I
> > believe putting an aggregation into an OAI-ORE format should be
> pretty
> > straight forward. The bigger question I have relates to what happens
> > when you have multiple different aggregations. How can you relate
> the
> > use of one piece of content in one aggregation to its use in another?
> > Do we end up with multiple ORE aggregations? Do we try to build an
> > uber aggregation? Is just storing individual relations in a
> > repository better than trying to maintain the whole (which might be
> > harder)? OAI-ORE is all about the aggregation, and doing things like
> > exchange, deposit, preservation, ... on the whole.
> >
> Yep, interesting question. Our proposal is focused, and I think
> correctly, on solving the problem we've encountered, which is that OER
> records in the form of those found within repositories such as Jorum
> are
> insufficiently described in terms of the constituent resources from
> which they are comprised - they are in fact an aggregation of resources
> and so should be described as such, for which purpose OAI-ORE is a good
> fit.
>
> How you'd describe the relations of a single resource to the
> aggregations of which it is part is a different matter. I wonder,
> actually, if there would be sufficient use cases for this to be a
> necessity, and whether to attempt to do so would actually then add
> complexity where none is needed. Certainly, it is not part of the
> specific need which we would intend to address.
It goes deeper than this though.
ORE is just one way of suggesting linked materials.
The Openlearn feeds include relevant resources using an enclosure tag - and in the Xpert database, that content is stored. Now if we could get into those PDFs and find pictures, we could provide those resources in the results too. Then a more formal method of understanding the relationship between content becomes really handy.
Jorum does a similar thing with a lot of pieces broken into parts, but then sadly no dc:relation or indication of associated pieces.
So there is lots of scope for providing more granular information on learning objects - and this would be great for a "remixing service".
|