ADMIRAL JISC Conference Demonstration Proposal
From ImageWeb
Contents |
Outline of proposed demonstration
The thrust of the demonstration will be to show the framework for desktop-to-repository acquisition and annotation of research data using simple, familiar, loosely coupled tools. The leading themes will be "sheer curation" and "curation by addition".
What we aim to demonstrate is that by engaging early with researchers, the acquisition and organization of information for publication need not be an arduous process, and that intelligent use of web interfaces allows us to assemble a complete front-to-back system from elements provided by disparate participants.
Sheer curation
We start from some arbitrary collection of researchers' data on their desktop or laptop computer. They are encouraged to maintain working copies on a shared file store because this gives them automatic daily backup and access to greater storage capacity; to use this, all they have to do is connect their computer to a file sharing area provided by ADMIRAL and copy their files there, all using tools with which they are already familiar.
(We might also show controlled web-based sharing with collaborators, but i feel that would be an unnecessary distraction for this demonstration.)
Preservation
For any data that is stored in the shared file store, we show a simple web interface that allows datasets to be selected and submitted to a University data repository service. The immediate benefit here for researchers is long-term preservation of the datasets. There is no constraint on the kind or format of data that can be submitted, other than possible overall size quotas that may be imposed. The submitted dataset content and metadata may be viewed using an regular web browser. There is provision for providing displays that are customized via ADMIRAL to reflect local needs (something that an institution-wide service cannot reasonably provide).
Curation by addition
If the local ADMIRAL data or metadata is updated or enlarged, it can be resubmitted and a new version is created in the Databank service. Thus, the preserved dataset can be incrementally improved until it is in a state fit for publication.
Metadata about the dataset is recorded simply in files stored in the ADMIRAL file store. Thus, new metadata can be created by a range of tools from simple text editors to sophisticated annotation tools. Metadata that is stored as RDF can be linked to the dataset manifest, and exposed through the Databank indexing and search services (((planned Databank capability))). (((Depending on the progress made in phase 2 of the project, we would aim to show some tools used to annotate the dataset with additional metadata.)))
Publication
Preserved datasets may be in a "dark archive" or a "light archive". Dark archive material is not intended for public view (e.g. personal information), and will not be further considered here.
Light archive material is associated with an embargo period, the default for which is (((TBD - researchers have suggested 5 years initially))). When the embargo period expires, the dataset becomes eligible for publication (((some details about authorizing publication to be worked out))). When the embargo period has expired and the dataset has been authorized for publication, it is made publicly visible on the web through the Databank service. We will demonstrate the public face of Databank for browsing and retrieving published datasets.
(((The embargo management is not yet implemented, but should be completed by the time of the demonstration.)))
(((Assuming progress as planned on the Databank service, we will also show how domain specific metadata can be exposed and searched through the Databank service, e.g. to locate images of gene expression of an identified gene in an identified organism.)))
Also, for data that is published, we will show that it can be identified, cited and accessed using a DOI. (((Not yet implemented, but should be available via Databank by the time of the demonstration. Assumes imminent deployment of British Library DOI issuing service.)))
Current implementation status
Apart from the elements noted, the demonstration elements described are all implemented and working (though in most cases needing some refinement of presentation).
