11/30/2011 - Hosted by Library of Congress

Cliff Lynch - Keynote

  • Repositories ("reps") gained traction in the 90s
  • Open Access movement is one origin of repositories ("author self-archiving") - Put journal articles in the institutional repository
  • Disciplinary repositories - Desire to circulate pre-prints as well as open access
  • In Europe, when articles were under copyright, authors put citations in the institutional repository
  • Coverage at most institutiions is pretty disappointing in absence of mandates; people argue that inst. reps have been a failure
  • Other view of Inst Reps:  Faculty now producing many artifacts that don't fall into category of text, journal articles. Repositories should provide a safe place for this material:  datasets, powerpoints, software, pre-prints, image collections, inputs to research, outputs, stuff used in teaching
  • Hard to measure success with this view; very different from goals of open access movement; no one can quantify how much of this material exists; little understanding of life-cycle patterns.  when is it appropriate the stuff into inst rep; short time-frame not adequate for measuring success
  • Thinking that non-institutional orgs should have repositories too:  NGOs, gov agencies, libraries, other non-profits; lot of room for development at this point, not much uptake
  • Substantial marketplace has emerged for inst reps: commercial and open source, repositories as a service; barriers are reduced
  • Many variations:  consortial reps, disciplinary reps.  
  • We are wrestling with the question of when it makes sense to do things on a disciplinary basis vs institutional basis; real advantages to disciplinary approach (sub-disciplines, special vocabularies, etc); but inst reps are the backstop right now; many disciplines will not be represented by repositories in near future; disciplinary reps tend to want to restrict to limited number of object types;
  • Challenges:  What sort of metadata should be included?  Conflicts between generality (apply to many cases) and specificity (allow easy discovery of specialized resources);
  • Metadata demands from libraries have made it a burden for faculty to deposit materials
  • Search engines tend to ignore third-party metadata; 
  • Aggregation of metadata across repositories is important
  • Need for sorting through author identity and cleaning up names
  • How should inst reps relate to storage of research data; are they appropriate for short-term storage of research data?  (research in high perf computing, eg); should there be a staging area before data is preserved?  we don't understand these questions very well;
  • Challenges for "small data" are just as great as big data
  • How do we move from a patchwork of inst and disciplinary reps into a network of repositories?  Some faculty have choice of both, shouldn't have to deposit twice; has been a challenging issue to migrate, though; Functional requirements of interoperable reps:  extract metadata (there is a standard); make automated deposit through a protocal interface; should be able to copy from one rep to another (different from a deposit?  Replication?  Object Reuse Exchange Protocol used for this but is complicated); 
  • Repository discovery and naming:  material needs to be accessible in long run; reps should assign unique, persistent IDs; how do you find reps?  You want to refer to repository at inst rather than URLs or specific hosts; we need registries of reps, lookup and discovery mechanisms
  • Major issue:  how do you cite data used in scholarly work, make reference to data in tables, how do we make correspondences between journal articles and data?
  • Institutional stewardship is a long-term commitment; they aren't always honored; repositores sometimes go away; stewards need to accommodate that reality
  • Questions:
  • Value of DataNet?  Cliff:  The projects are capacity building and integrative, linking repositories together, providing tools to act on them; 
  • Lots of distributed efforts, where should the focus be?  Cliff:  It's complex:  many scholarly communities are stakeholders, public uses the materials; Difficult to pull everything together while respecting specialization; OR conference helps bring people together; DataCite, ORCID are helpful efforts; CNI tries to provide a home, National Academies tries to pull together science communities; international data curation conference has been good venue; Chief Research Officers at universities don't seem to have a meeting like other university execs

Case Studies- Jerry Sheehan

  • Delivering Data in Science (March) - in Paris
  • #stirepos is today's hashtag
  • Jane Greeberg - UNC Chapel Hill (Dryad)
  • Dryad is a collaborative, run by a consortium of journals
  • Objectives:  repository for research underlying peer-reviewed publications in basic and applied sciences
  • Partnership with journals, which have a data archiving policy
  • Dryad associated with DataOne
  • Data built on DSpace; work with @mire, "the company that oversees the DSpace software"
  • Federated searching with TreeBASE and KNB LTER
  • Dr. Ian Bruno - Cambridge Crystallographic Data Cetner
  • study of molecules - use in drug design and development
  • 140 industrial subscribers sustain their efforts
  • Sustainability is still an issue; big pharma has been impacted financially; have competition with commercial apps
  • Fuzziness over where ownership rests
  • Value is added for subscription service; resentment that data is not open
  • H.K. Ramapriyan, Earth Science Data and Information Systems Project, NASA Goddard Space Flight Center - EOSDIS
  • Earth observing satellites and earth science measurements
  • Mission is to meet the challenges of climate and environmental change
  • Data is available at no cost; EOSDIS provides data processing, management, interoperable data archives
  • Satellite data is captured by flight operations, processed, sent to multiple data centers
  • Other sources of data too; multiple types of instruments 
  • Middleware and associated clients provide search and access to data across al data centers
  • Distributed data centers handle different types of data (e.g., National Snow & Ice Data Center)
  • There is global directory; all datasets are discoverable; cross data center searches through REVERB
  • Many data visualization and analysis tools
  • 5.1 petabytes of data
  • DSpace@MIT
  • PubMed Central
  • Electronic extension of NLM's print journal archive
  • Free access; 
  • Deposit Paths:  publisher sends final article in XML or author sends manuscript file, it's processed and NIH creates XML, then deposited
  • NLM has formal agreements with publishers (final copy, deposits are permanent, publishers can't withdaw content
  • They have non-exclusive license to use the content; they don't own it
  • Author must retain rights to manuscript before signing publication agreement
  • PubMed DTD now a NISO standard
  • Library of Congress
  • LOC is a holder of large datasets that are used in research (e.g., Twitter)
  • Mandatory Copyright Deposit now bringing in many new files
  • So far their system is discovery and delivery; lacking many repository features
  • DataCite and EZID
  • Creates a global citation framework for data
  • Uses DLI (Digital Logic Identifier)
  • Take a lifecycle approach www.cdlib.org
  • UC3DCXL - open source add-in for Microsoft Excel as a data collection tool 
  • n2t.net/ezid (create an ID)
  • ORCID - Brian Wilson, Thomson Reuters
  • Open Researcher and Contributor ID
  • Allows reliable attribution of authors and contributors
  • ORCID allows you to create a profile associated with your ID
  • 282 participating orgs internationally; academic, publishers, government, societies, non-profits, etc.
  • Mellon granted award to MIT, Harwaverd, Cornell to study ORCID business models
  • VIVO awarded grant to ORCID for collaborative research
  • Just released first code
  • Will hire executive director and technical director
  • Institutional seeding of profiles, delegated management
  • To use OAuth (used by Google, Facebook)
  • self asserted, socially validated, organizationally asserted identity = more credible assertion
  • Chris Greer - NIST
  • Promte infrastructures as well as standards; consider themselves part of the "data community"
  • Missing:  a framework for the community to make decisions; 
  • Draws comparison to NISTs mandate to design interoperable smart grid; information requirements are similar; many different stakeholders; Smart Grid Interoperability Panel - consensus based organization; 724 members; architecture committee, testing and validation, security; stakeholders include standards bodies, regulators; participation is voluntary, but you must participate -- miss sequential mtgs or votes and you're voted off the island
  • If the data community did this, NIST would be the convener, would have White House support
  •  
  • No labels

All content on the LYRASIS Wiki is licensed under the CC BY (Attribution) license, unless otherwise noted.