Repositories at ASIS(&T)

Looking over my notes, I have a lot more to say about the Institutional Repository program than I wrote earlier.

The most relevant for us here in duckland is Harvard's. The Harvard IR is for the science libraries, and its mission is explicitly to support changes in scholarly publishing—nice to see the big girls putting their money behind a worthwhile effort. It’s also an experimental system: they are intentionally scaling up to see if D-Space and the libraries can handle it.

They also building interoperability features. They want to play nicely with oai-pmh and open-url type services. And…they're putting in serials!! Just a few weeks ago we were speculating whether or not D-space could serve as an e-journal publishing platform.

The BEST part of what Harvard is doing is in policy and implementation. (Well, I don’t know if they’re the best, but I found them the most inspiring.) They are working from the notion that faculty are overloaded and that they don't want any additional work to be added to the publishing process. Their job is to get their papers in a peer review journal. Period. Therefore convenience and ease-of-use have become driving factors. One concrete result of this philosophy: the librarians are going out and determining which researchers have published papers in journals that allow self-archiving. They then contact the researcher and ask him/her to allow the library to archive the work. The only thing needed from the researcher is permission. The library does all the labor of getting the paper into the IR.

Another interesting aspect of Harvard is that they are not limiting the repository to PDFs and word documents. When Harvard talks about types of materials, they are talking about genres—not filetypes. Consequently, they are accepting scientific datasets into their IR. This has led to a metadata problem: DCQ doesn’t work very well for data—therefore they are looking at DDI.

That said, if you follow the link over there, they don’t have a big chunk of content posted yet. Hopefully in the coming weeks we’ll see it opened up a bit.

The University of Virginia appears to have gone nuts with building their own system. Not really an IR, this one seems to be more along the lines of a ContentDM + text based documents digital library collection. This is a Fedora localization—I don’t know squat about Fedora, but it seems to be a metadata management platform a la ContentDM—but without the public interface. They built the public side on their own, and are migrating 100,000 images from a previous system into this new one.

A remarkable thing about this project is that they had an actual requirements analysis and design process—they didn’t just close the office door and pound it out. They talked about how they wanted the system to work and then mocked up the functions screen by screen. They’re image viewer is sweet—-it actually has contrast and brightness controls.

The other thing to note here is that Fedora is the repository—the interface is home grown and the search engine is a 3rd party system. I hope it was worth the effort.

MIT, the originator of D-Space, has some unique things going with their program. They have digitized all the out-of-print MIT Press books and allow on-campus access to the whole catalog. They will also be allowing off-campus viewing of MIT theses, but will maintain a fee-based printing service.

Like Harvard, MIT is being very liberal with filetypes—but is issuing strong warnings about preservation. They are also offering a ‘premium’ repository service where they will take on tasks such as metadata entry, copyright clearance, and data migration for a fee. The guy in charge of the premium service had drinks with us one evening—he seemed a little nervous about being a campus entrepreneur, but he also seemed completely capable of pulling it off. Definitely a project to watch.