I’ve talked about this many times before, but this is probably the first time that I’ve committed pen to paper (or, keypress to web-page, in this Web 2.0 world) and described it publicly.
I believe that the current Repository concept is flawed… actually, lots of people now realise that the current concept is flawed, and people are trying to decide what the correct model should be.
We know that the binary-object-centric view does not work, and we know that we need to involve the user much much earlier in the aquisition process
Here is my model:
Researchers have Interests: a general topic/area/subject that interests them; that they want to investigate; to understand.
Researchers apply for funds to Research aspects of that Interest: to look at particular facets; to search for significance; to find reasons or rules of behaviour.
Researchers produce things: they write articles; they produce data; they have significant emails; they go to conferences…. and all of these relate to specific pieces of Research, or to general Interests.
Articles go through iterations: a lengthy final draft; a pre-submission version tailored to a specific journal; a post-review version; a publishers final-copy version… and each of these link to the previous version, or directly to a specific Research thing or Interest thing
The idea is that when someone puts a thing into their workspace, they describe it:
- An Interest will have a Description, and inherit the primary “author” from the user profile.
- A piece of Research will have a Title, a Description, a Funder (and “grant number” or other code), will default the Principle Investigator to be Interest.Author.
- The various Article/Conference/Email/Data things will have their appropriate meta-data, and inherit from the item above them..
The idea, therefore, is that the researcher deposits often, and rarely has to provide much supportive metadata… which could easily fit into their normal working flow… and if we could promise to back up their data, they will probably be happy.
Look at http://www.myexperiment.org/ as an example of something heading in this direction.
 If we are clever, then we can link into the organisations MIS databases, and pull the “principle investigator” from there, and get a list of assocciated researchers (those paid from the grant), who are likely to be co-authors.