This presentation (the last of the day) by Sam Pepler from a data centre.

The OJIMS project is:

• Overlay Journal Infrastructure for Meteorological Science
• JISC and NERC funded
• Looking specifically at a ‘data journal’ rather than traditional publication
• Looking to evaluate business models for overlay journals
• Creating an open access subject based repository for meteorology etc.

The University of Leeds will lead development of a dataset review policy with the Royal Meteorological Socitey (RMetS).

A ‘data journal’ is a journal that links published documents with the data that the publication uses – cf CLADDIER another JISC funded project.

What are the benefits of a data journal?

• Extend the value of peer-review from papers to data, to provide assurance that data documentation meets the necessary scientific standards
  • Metadata standards
  • Independently understandable
  • Re-useable
  • N.B. about quality of data documentation – not about quality of data set (i.e. “can you use it”, not “is it useful”)
• Provide an overview of the quality and applications of data, enabling it to be used more easily and appropriately in research and applications
  • adding independent quality statements about usefulness
• Provide recognition of the work of collecting and describing data
  • High quality, reusable data is not presently a citable resources
  • The writers of papers do not necessarily acknowledge those who collected the data

Why make an overlay journal?

• Data already in ‘a repository’ – just needs some independent review
• Because data is bulky, compound and complex – not easy to copy (possibly not as ‘self contained’ as traditional published paper?)

MetRep is a subject based repository for meteorological sciences. This was seen as filling a gap in the market – there is no store for some of the items they want to store. Examples of MetRep items are:

• Paper from ‘Weather’
• Set of pictures illustrating cloud forms (e.g. teaching aid)
• Report documenting a file format for climate models
• Weather balloon data
• Recording of a interview with ministers about climate change
• IPCC reports
• Logo for a research programme

Although some of these could sit in existing repositories – Institutional Repository, JORUM, websites, etc.

Perhaps MetRep should be an overlay repository? What does it mean to say an item is ‘in the repository’?

So – what is proposed is:

• Establishing a ‘Overlay document’
  • Metadata about the overlay document
  • Review process information
  • Discovery metadata for the reference document
  • Reference to document (referenceable via a resolvable id in a trusted repository)
• The ‘review process information’ consists of
  • Version of document in review cycle
    • Submitted
    • In review
    • Published
  • Public comments
  • Description of review process
  • Digital signature?
• Metadata about the overlay document would contain
  • Author (of overlay no the referenced document)
  • Other DC (Dublin Core) fields
• Discovery metadata for the referenced document and Reference to document
  • DC metadata harvested from document (not sure if he means from the document, or from the metadata associated with the document?)
  • Resolvable reference to document
  • Other identifiers for document

The overlay repository would have overlay documents pointing to both documents or data

The advantages he sees in this approach:

• Clear the ‘overlay’ is a document about another document – the two items are distinct self contained
• Authorship for the referenced and referencing document are allowed to be different – others can submit a document for review
• The overlay document has the same meaning as a stand alone item – you can take it out of the repository context, and is still meaningful
• Review mechanisms and repositories do not need adapting to deal with these items
• You can review a private document/data set – answers the ‘is thing worth buying?’ question

Disadvantages:

• Authentication issues – might be able to ‘fake’ items?
• What if the author does not wish for document to be reviewed?

Implementation:

• Atom XML representation (mention of OAI-ORE here)
• Already a popular format with many tools
• Need a tool to create the records
• Need a web rendering method

Trusting repositories:

• More than resolvable identifiers – need to believe the object is preserved
• Need to know what preservation means for complex objects
• Repositories need to have sound footing – but there are no absolute guarantees

Somewhere along the line I’ve lost the point of what we are trying to achieve with this approach – Sam is now summarising, so hopefully this will help:

• OJIMS is about widening review processes beyond papers
• This means storing a wider range of objects – hence MetRep
• Data is a good e.g. of valued stuff which is not recognised in formal manner – hence ‘data journal’
• Lots of repositories are already storing the things – hence ‘overlay repository’
• … didn’t get the last couple of points

Overall seems to be about a way of recording the ‘review process’ alongside the actual object being reviewed.