What is Data Anyway? Findings from the StORe Project

1
What is Data Anyway? Findings from the StORe
Project

• John MacCollEdinburgh University Library

2
Structure

1. Thoughts about data and business models
2. StORe and its findings
3. Curators, publishing and libraries emerging
   roles and models

3
The bioscientist's question

• Need we polarise (source and outputs)?
• Prose is problematic
• Structured data publication is more accurate
• Relying on the literature Google-ises access to
  results
• It is not scientific

4
Instead ...

• Mark-up statistical results
• (i.e. enrich the data)
• Publish directly on the web in XML databases
• Link to commonly agreed domain ontologies
• (Publish the papers as well)

5
But

• Where is the business model?
• Who pays?
• Who does the curation?
• Libraries?

6
(No Transcript)
7
The Survey
8
Two-Way Links?

• 85 support project aims as potentially
  advantageous to conduct of research
• Would save so much time, making research more
  productive
• Would allow reanalysis as new methods emerged
• Integration of multiple data sets from different
  publications

9
Open Access?

• It should be a requirement that data from
  publicly funded research is freely available
• Restrict access until results are published to
  prevent data scavenging
• A creditable aspiration but without a data
  administrator this represents a large burden from
  editing, compiling and sanctioning release

10
Data management?

• 75 generate and use complex data sets
• Storage of unique and original research on PCs
  and laptops is commonplace
• Access influenced by perceived absence of
  adequate protection and need for interpretation
• Data is held on secured CDs in encrypted format
  with only an identifying code. The codebook is
  kept physically separate.
• Data volume and lack of time/experience compound
  issues of data ownership
• Im not encouraged or discouraged from providing
  data. It just does not justify the effort.

11
Repositories?

• Development and use culture of self-sufficiency
  broad range of effectiveness
• Some sophistication but limited understanding and
  familiarity also found across the sector
• Very happy with what we have in astronomyplease
  dont mess with them for the sake of some
  aesthetic..
• ... my understanding of this topic is so limited
  that giving any answer would be trivial

12
Metadata?

• Appropriate assignment acknowledged to be
• Critical
• Demanding (intellectually and in the time it
  requires)
• Consensus on the need for good metadata does not
  necessarily translate into its provision
• High level of self-assignment

13
Scientists and Metadata the Facts

• I decide which terms to use and I assign
  them 212
• Research colleagues assign metadata on the team's
  behalf 55
• Research support staff assign metadata on the
  team's behalf 22
• Metadata are assigned by library/information
  services staff 4
• Metadata are assigned by the repository
  administrators 37
• Metadata are generated automatically 63
• It is not known who assigns metadata 68
• Other (please specify) 37

14
Support?

• Lack of awareness of available support
• Disinclination to seek support
• Self-reliance with IT matters
• Prefer online or documentary support
• Discipline knowledge essential
• With few exceptions, management of research data
  not usually associated with librarians
• Yet

15

• Qualified demonstration of expertise in metadata,
  data preservation or curation
• Examples of uncritical use of sources
• Both access and sharing restricted by lack of
  confidence in processes
• Browsing, online help and other features need
  expert support and maintenance
• Need to boost awareness of opportunities from
  electronic data management

16
Corroboration

• Researchers and discovery services behaviour,
  perceptions and needs RIN report, November 2006
• contact with librarians and information
  professionals is rare
• researchers are generally confident in their
  self-taught abilities.., librarians see them
  as..relatively unsophisticated
• librarians see it as a problem that they are
  not reaching all researchers with formal
  training, whereas most researchers dont think
  they need it

17
Improving the Research Management Lifecycle

• Hypothesis
• Undertake research/experiment
• Produce data/commence data curation
• Publish data ? and/or paper? - bioscientist
• Select and organise evidence
• Write/assemble paper
• Submit to peer review
• Apply revisions and produce final draft
• Publish paper ?
• Activate data-publication links
• Peer review of data?

18
Two-way links - benefits and risks

• Opportunities to
• Explore a deeper level of detail
• Validate experiments
• Track the use and improvement of research output
• Identify collaborators
• Confirm completeness of information searches
• Supplement published papers

• Potential risks from
• Uncertainty of peer review
• Premature dissemination
• Subversion of scholarly paper
• Scavenging
• Lack of interpretive data

19
The StORe Pilot Demonstrator

• Allows publication only if data has been
  deposited or identified
• Groups items (data and publications) as projects
• Accepts projects based on either primary or
  secondary analysis
• Primary analysis creates new data
• Secondary analysis is based on existing data and
  may or may not create new data
• Source repository - the UK Data Archive
• Output repository - the LSEs Research Articles
  Online
• Pilot federation - includes a test institutional
  repository at the University of Essex

20
(No Transcript)
21
How StORe works

• At present, search is across metadata in Essex
• Find an article and you can
• find the associated data
• move to the official versions at LSE or UKDA
• list all articles and related items of the author
• Find data and you can
• find all articles associated with them
• e-mail data owner to request access to approved
  but embargoed/private data

22
How StORe works

• A user can search across all or specific
  collections without being logged in, using a
  simple Google-type search

23

• ... or by employing more advanced options

24

• Registered users can also view data within
• Solely owned private or public collections
• Collaborative collections within a federated
  source or output repository
• Collections to which they are a contributor

25
StORe the future?

• Essex as multi-disciplined institutional
  repository in large federation of
• source output repositories plus
• StORe-enabled institutional repositories at other
  HE/FE institutions
• If all universities had a StORe system,
  collaborations could be established with the
  institutional repository/repositories, and all
  relevant data repositories. Researchers would use
  StORe as their single route to deposit and
  associate
• They would need curators to assist

26
StORe one approach to data publishing

• Data deposit in institutional repositories until
  accepted by source repository
• Source repository verifies data authenticity
• Publication dependent on data deposit and
  subject to output repository controls
• Access to non-public objects can be restricted
  and requires authentication
• Release of data can be embargoed
• Curators should know about it and its alternatives

27
Curators link across paradigms

• Curation embraces appraisal, description,
  preservation, rights management ...
• It requires understanding of science at its
  domain levels
• Good curation is key to
• achieving more rapid discovery
• achieving more efficient research
• preventing bad practice
• preventing lazy supervision
• preventing sloppy analysis
• preventing fraudulent claims

28
Scientific Publication

• As prose
• Journals provided good practice for a long time
• Now no longer true, hence reinvention (Open
  Access, Creative Commons, SPARC, etc)
• There was a business model which is now broken
• As data
• Has never existed
• Is beginning to be invented
• There is no business model

29
Libraries?

• Exist to permit sharing
• Protect fragile business models
• Sustain otherwise uneconomic enterprises
• Financial role will be to pay for publication
  embracing prose and data, and subsuming curation
• Must help researchers to publish their data
• Because the prose-data hierarchy is being
  gradually subverted (and peer review will
  ulimately require it)

30
Thank You