DSpace Status Report, 3/21/2006

We have four different types of content in our repository:

1. electronic theses and dissertations
2. images
3. 'excellent undergraduate research"
4. "grey" literature authored by faculty
   
   

#1 uses the ETD-db software from Virginia Tech. #2 begins its life in a Filemaker Pro database and gets exported to DigiTool. #3 and #4 both live in DSpace. For the most part, each type of content is described using Dublin Core and each type of content is harvested via OAI into a centralized (MyLibrary) cache.

We have successfully written reports against the cache allowing us to provide (SRU) searchable as well as browsable interfaces to the content. We have been able to create searchable as well as browsable interfaces to subsets of the cache too. We have been able to create"widgets, gadgets, and gizmos" against the cache to provide value- added services such as the syndication of ordered lists of content to departmental and author pages. I have also demonstrated how pages crawled by Google can be listed with their associated Google PageRank integer denoting a sort of impact factor.

We have successfully drafted a release/permissions form. It stipulates four things:

1. Yes, I am the copyright owner of this work.
2. I grant the University to deposit my work in the repository and freely distribute it.
3. I retain my copyrights to the work.
4. I may pull my work from the repository as along as I submit my request in writing.

To date we have less than 750 items in our repository. Half of them are images. The other half are working papers from two on-campus institutes and a tiny number of undergraduate papers. The balance are theses and dissertations. In the near future I believe we will include about 100 computer science technical reports and about 100 citations from the engineering faculty. We hope to include content from the Open Course Ware (OCW) project as well.

When it comes to authority control we have not done a whole lot. To implement a controlled vocabulary we have used our facet/term approach to classification. We created a new DSpace/Dublin Core field called description.ft. In it we place our facet/term combinations, and they look something like this:

  \\Subjects\\Latin America
  \\Subjects\\International Studies
  \\Flags\\Undergraduate Research
  \\Regions\\Mexico
  \\Regions\\Venezuela

When we harvest this data we take note of the incoming data's shape (\\facet\\term). This way we know it is a facet/term combination. We then update our cache accordingly. We need to implement these "shapes" because Dublin Core data harvested from DSpace is unqualified; everything comes out as flat title, creator, description, etc.

Just today we articulated a way to exploit our facet/term approach to implement an authority list. We will create a new facet called Bio. We will then create terms using University members' campus ID's as terms. (Mine is emorgan.) These terms will then be associated in our MyLibrary database with these (Dublin Core) fields:

  * title will hold the name of a person
  * description will hold the person's bio

In our DSpace implementation we then enter appropriate facet/term combinations into bibliographic records so we will add things like \\Bio\\emorgan to records as well as the facet/term combinations above. While harvesting we will "join" bibliographic records with biographic records to display biographies along side working papers. In the future we may hack the DSpace data-entry interface to allow the selection of these Bio facets via a pull-down menu.

That is what we have done to date. We are progressing. I'm sure there is something I've left out, but...  HTH, and again, thanks for asking.