Tim Sigmon is talking about the development of the UVa digital library, where they area attempting to really offer integrated searching and delivery of digital library content.
One of the issues was coming up with common metadata for describing these digital objects. There was a steering group to review formats and come up with standards. There are descriptive and administrative metadata standards.
They also needed new specs for how images would be stored in these collections. Three content models were developed uvaHighRes, which includes preview, screen-sized and high quality large image; ivaLowRes – only preview and screen-sized images; and uvaBitonal- bitonal TIFFs only. One content model and production standard were set for image metadata.
Texts are represented in a local extension of the TEI DTD, along with encoding guudelines. There are three content models for text: uvaGenText – transcription with no page images; uvaPageBook – page images with no transcription; and uvaBook which has both transcription and page images. All page images must conform to the image standards.
Archival finding aids were specified – uvaEAD (encoded archival description 2002). Images and texts must conform to the content models.
There are two default disseminators on every object: Default access behavior, including getPreview, getFullView, getLabel, getDefaultContent; and Admin and descriptive metadata behaviors. There are also class-specific disseminators for different kinds of objects.
They built some tools for users of the DL, including a “shopping cart” for people to collect their digital objects as they search and browse, and then do things with that collection, e.g. to create a slide-show for a lecture.
They had to create processes to convert legacy images, texts, and finding aids, as well as the work flow for getting content into the repository – this wasn’t primarily a technical issue but changing the way library catalogers do business.
There’s a demo at http://www.lib.virginia.edu/digital/collections/