[Some quick'n'dirty updates as this slightly re-purposed wiki has sprung back into life in 2009]
This Museums and the Web paper is a useful reference: Sharing cultural heritage the linked open data way: why you should sign up.
Make your Museum API, part 1: Ruby. Advice on getting started.
Read more: Sharing cultural heritage the linked open data way: why you should sign up | museumsandtheweb.com
Ideally real-world, but possibly hypothetical use cases and user stories that relate to real problems that those in the intended audience might have. I'm not sure what format they should take, but for the moment a simple statement of the type of user (curator, exhibition staff, educator, etc), their goal or problem, and a possible solution should do. They could be grouped into headings by type of issue or left in a more free-form state for now.
The Research Information Network has a report, Discovering objects: Meeting researchers’ needs, which could provide some good examples. It "investigates how researchers discover and gain access to physical objects and artefacts using four varied subject disciplines as examples of the process: archaeology, art history, earth sciences, and social and economic history". In summary:
"The report finds that researchers want access to online finding aids to enable them to plan their visits to museums and collections, so that they can both see and handle the objects, and that contact with curatorial staff is of critical importance. The evidence gathered shows that most researchers are unaware of the online catalogues that currently exist or are being developed and feel that that there is a lack of consistency and transparency in the arrangements that museums make for researchers’ direct access to objects. However, their most important wish is that online access to museum databases to be provided as quickly as possible, even if the records are imperfect or incomplete."
The user: the curator of an online exhibition in Museum A needs to incorporate 'loan' records from Museum B. They can be recorded as loan objects in the Collections Management System.
The problem: how can she obtain the relevant records from Museum B and incorporate them in her online project without re-keying the data from a CSV export from Museum B?
Museum A has content hosted on an image licensing site. The licensing site uses its own internal identifiers and doesn't display Museum A's accession numbers or internal collections management system identifiers.
Museum A would like to direct visitors from certain collections pages to page with images of that object on the licensing site but there's no link between the Museum identifier and the licensing site identifier.
Solution A: a concordance table, stored in a database on the museum site and used to generate links to the licensing site. However, this quickly becomes out of date.
Solution B: an API on the licensing site that lets Museum A send their accession number and get back the licensing site identifier.
(i.e. using your own APIs and linked data services internally)
BBC Sport website refresh: a new look and consistent design
"Journalists tag BBC Sport’s content objects such as a story or a video with the team name or competition they are relevant to.
The linked data stores these semantic tags and a generic metadata model about each piece of content, providing an API to allow the retrieval of relevant content about each team or competition that is then dynamically aggregated in a PHP-based Page Assembly Layer (PAL), the Sport PAL app.
We used the same application to aggregate content around ‘smaller sports’ like taekwondo, fencing and ice hockey, again making use of linked data.
As both areas of the site were in our Sport PAL app we were able to share page components with the rest of the site."
[wrong format, clipped from EDL presentation, sorry:]