The following is the official press release describing the proceedings
of the UPS Initiative's very important meeting last week in Santa Fe.
It is followed at the very end by some unofficial addenda by me,
particularly about how the newly agreed upon Santa Fe standards will be
applied to new, generic archiving software that will be created here at
Southampton in the next 6 months and will then be given away free to
all universities worldwide who wish to establish Eprint Archives for
the research papers of all their Faculty, with sectors devoted to each
of their academic discipline.
(The name "UPS" will shortly be changed to "Open Archives Initiative" to
reflect the fact that the initiative is decidedly not just about
"Preprints" but about creating interoperable archives for "Eprints,"
which includes unrefereed, unpublished preprints, refereed, published
reprints, and related kinds of research documents and data.)
------------------------------------------------------------------------
First meeting of the Universal Preprint Service Initiative
UPS Initiative: Paul Ginsparg, Rick Luce, Herbert Van de Sompel
<http://vole.lanl.gov/ups/ups1-press.htm>
Meeting:
* Location: Santa Fe, New Mexcio, US, October 21-22
1999
* Sponsors: Council on Library and Information
Resources, the Digital Library Federation, the
Scholarly Publishing and Academic Resources
Coalition, Association of Research Libraries, the
Research Library of the Los Alamos National
Laboratory.
* Meeting moderators: Clifford Lynch & Don Waters.
* Represented institutions/organizations: American
Physical Society, Andrew W. Mellon Foundation,
Association of Research Libraries, California
Institute of Technology, Coalition for Networked
Information, Cornell University, Council on
Library and Information Resources, Digital
Library Federation, Harvard University, HighWire
Press, Library of Congress, Los Alamos National
Laboratory, Massachusetts Institute of
Technology, NASA Langley, Old Dominion
University, the Scholarly Publishing and Academic
Resources Coalition, Stanford Linear Accelerator
Center, University of California, University of
Ghent, University of Southampton, University of
Surrey, Vanderbilt University, Virginia Tech and
Washington University.
* Represented eprint-initiatives: arXiv.org (=xxx),
CogPrints, NDLTD, RePEc, EconWPA, NCSTRL, NTRS
* Participants: see seperate list
Executive Summary
The Universal Preprint Service initiative has been set up to create a
forum to discuss and solve matters of interoperability between author
self-archiving solutions, as a way to promote their global acceptance
(see http://vole.lanl.gov/ups/ups.htm ).
The first, largest and most important such archive is the Los Alamos
National Laboratory (LANL) Physics Archive. Founded by Paul Ginsparg in
1991, LANL now houses over 100,000 papers, mirrored worldwide in 15
countries with over 50,000 users daily and still growing (see
http://arXiv.org/cgi-bin/show_stats ). Other disciplines and
institutions have begun to create public research archives along the
lines of LANL but what is needed are conventions that archives could
adopt to ensure that they work together so that any paper in any of
these archives could be found from anyone's desktop worldwide, as if it
were all in one virtual public library.
The participants in the meeting were digital librarians and computer
scientists specializing in archiving, metadata, and interoperability,
and they included the founders of the principal public research
archives that exist so far. The participants were diverse in their
underlying motivations, but entirely unified in their objective of
paving the way for universal public archiving of the scientific and
scholarly research literature on the Web.
The group agreed on minimal technical requirements for archives. These
will be published seperately as the "Santa Fe Conventions" and, in the
next six months, will be implemented in the existing archives.
Technical Summary
The first meeting concentrated on the creation of cross-archive
end-user services. The aim was to try and identify general
architectural and technical characteristics of archive solutions, that
would facilitate the creation of such services. These characteristics
could then be used as recommendations for existing and upcoming
initiatives.
The meeting started off with a presentation and demonstration by a team
consisting of Herbert Van de Sompel (University of Ghent and Los Alamos
National Laboratory), Michael Nelson (NASA Langley and Old Dominion
University) and Thomas Krichel (University of Surrey and RePEc
initiative). This group had built an experimental end-user service
providing access to data originating from main archive initiatives
(arXiv, RePEc, NCSTRL, NDLTD, NTRS). A variety of technologies were
used in the project, including NCSTRL+ as the digital library service,
intelligent objects called buckets as a means to store the archive
metadata and the SFX linking solution as a means to interlink the
eprint data with the traditional scholarly communication mechanism. The
presentation identified problems that arose during the project, and
discussion of those served to launch the UPS meeting. This presentation
was followed by position papers on interoperability issues presented by
Carl Lagoze (Cornell University), Kurt Maly (Old Dominion University),
Ed Fox (Virginia Tech) and Carolyne Arms (Library of Congress).
Following the initial presentations, there was a panel discussion in
which Paul Ginsparg (Los Alamos National Laboratory), Paul Gherman
(Vanderbilt University), Eric Van de Velde (CalTech) and John Ober
(University of California) expressed their opinion on the possible pros
and cons of institutional versus discipline-oriented archive
initiatives. The UPS group concluded that many different archive
initiatives were likely to emerge, with different conceptual,
organizational and technical foundations. In order for such initiatives
to successfully become part of the scholarly communication system,
interoperability was seen as a crucial factor.
The UPS group agreed that interoperability hinges on a fundamental
distinction between the archive-functions, which include
data-collection and maintenance and end-user functions, like the
cross-system search and linking prototype service described in the
opening session. Although archive initiatives can implement their own
end-user services, it is essential that the archives remain "open" in
order to allow others to equally create such services. This concept was
formalized in the distinction between providers of data (the archive
initiatives) and implementers of data services (the initiatives that
want to create end-user services for archive initiatives).
Stimulated by a presentation by Thomas Krichel, the UPS group agreed that an
essential feature of the Santa Fe Conventions would be that providers of
data use a standard mechanism to state the conditions under which their
datasets can be used by implementers of data services. Similarly, the
implementers of data services could describe the use they make of archive
data.
This organizational argument was followed by a discussion on the
technicalities of creating end-user services for data originating from
different archives. The group recognized that there are basically two
ways to implement these: a distributed searching approach and a
harvesting approach. The former would require archives to implement a
joint distributed search protocol, which is not considered to be a
low-entry requirement. Moreover, the technical experts recognized that
there are important problems of scale when implementing such
distributed search solutions, in light of the possible emergence of
thousands of institutional and/or subject-oriented archives worldwide.
As such, the group decided this was not a realistic approach at this
point in time. Therefore, as in the experimental project presented at
the beginning of the meeting, a harvesting solution was proposed. Such
a harvesting solution would allow trusted parties - the ones that
subscribe to the Santa Fe Conventions - to selectively collect data
from different archives. It was identified that such a technique
requires an understanding regarding:
* Protocols to selectively harvest data;
* Criteria that can be used to selectively harvest data;
* Metadata formats that are used by archive solutions to respond to
harvesting requests.
It was recognized that providers of data could describe the details of these
interfaces in standard ways thus enabling implementers of data to create
archive-specific harvesters. Still, the UPS group decided to go one step
further and to highly recommend the following:
* Protocols to selectively harvest data: implementation of part of the
Dienst protocol in order to achieve a uniform way to poll an archive
for its logical division(s) (subarchives) and to selectively harvest
data from these divisions.
* Criteria that can be used to selectively harvest data: there should at
least be support for a bulk harvest of all data from an archive, as
well as a mechanism to harvest based on accession date. Other
harvesting criteria that were thought to be important included author
affiliation, subject, publication type.
* Metadata formats that are used by archive solutions to respond to
harvesting requests: It is recognized that archives will use (an)
internal metadata format(s) best suited to deal with the material to be
described. Still, the UPS group decided to propose a minimal Dublin
Core compliant metadata set, called the Santa Fe Set, that should be
made available by all archives. It is desirable that archives are able
to respond to harvesting requests with data delivered in both the
internal metadata format as in the Santa Fe Set format.
The representatives of existing archive initiatives at the meeting as
well as those from institutions that are in the process of setting up
archive initiatives agreed to comply to those guidelines. The Dienst
protocol will be enhanced to allow for the functions mentioned above
and a minimal Dienst release facilitating the process of making an
archive compliant to the required aspects of Dienst will be made
available. A transport format for MARC-formatted metadata will be
proposed, as well as an XML DTD for the description of the Santa Fe
Set. The recommendations will be extensively documented on a Web site.
Adoption of the recommendations will be promoted worldwide.
The way forward
* The minimal Dienst protocol set will be implemented for all archives
that were represented at the meeting. This will allow for a first round
of experimentation with the creation of end-user services layered over
existing archives.
* The group identified the urgent need to discuss the mechanisms used to
submit material to archives.
* Paul Ginsparg suggested that a next meeting should be held in Europe,
in the first quarter of next year.
* It was also thought to be important to have a presentation and/or
workshop on the UPS Initiative at the ACM 2000 Conference on Digital
Libraries as well as at the European ECDLC.
* The experimental, non-productional prototype that was presented at the
meeting will temporarily be available for exploration at the beginning
of November 1999 at http://ups.cs.odu.edu . The representatives of Old
Dominion University, the Research Library of the Los Alamos National
Laboratory and the University of Ghent expressed their interest in
continuing this prototyping work.
* The UPS Initiative will soon be given a new name and Web site.
__________________
October 29th 1999
get in touch with the UPS
initiative by contacting
herbert.vandesompel@rug.ac.be
------------------------------------------------------------
Unofficial addenda and application to generalization of CogPrints
Archive to all academic disciplines (Stevan Harnad):
The Santa Fe meeting was about public archiving of scientific and
scholarly research on the Web. <http://vole.lanl.gov/ups/ups.htm>
What follows is my own (nonofficial) summary; first, the context:
The first, largest and most important such archive is the Los Alamos
National Laboratory (LANL) Physics Archive. Founded by Paul Ginsparg in
1991, LANL now houses over 100, 000 papers, mirrored worldwide in 15
countries with over 50,000 users daily and still growing.
<http://xxx.lanl.gov/cgi-bin/show_monthly_submissions>
Other disciplines and institutions have begun to found public research
archives along the lines of LANL but what was needed to promote this
public archiving intitiative was conventions that could be jointly
adopted to ensure that the archives will be mutually "interoperable,"
which means that they can be integrated seamlessly into one globally
navigable archive (so one need not know where to look for what in
advance). With suitable "metadata" tagging that they all share (for
example, by title, author, subject, date), any paper in any of these
archives could be found from anyone's desktop worldwide, as if it were
all in one virtual public library.
The participants in the meeting were digital librarians and computer
scientists specializing in archiving, metadata, and interoperability,
and they included the founders of the principal public research
archives that exist so far. Standards and protocols were agreed upon,
and in the next six months these will be implemented in the existing
archives, and put forward with a name something like "The Santa Fe
Agreement," recommended for adoption globally.
The Santa Fe participants were diverse in their underlying motivations,
but entirely unified in their objective of paving the way for universal
public archiving of the scientific and scholarly research literature on
the Web. All agreed that this literature is currently being held
hostage and needs to be freed. All wanted to free it from (1) the
access barriers of the paper medium; most wanted also to free it from
(2) the access barriers of journal subscription prices; some wanted to
go on and free it also from (3) the access barriers of journal peer
review. But the necessary condition for any of these is an
interoperable digital literature; the Santa Fe protocol should go a
long way toward bringing that to pass.
For my own part, I am a confirmed advocate of (1) and (2) and an
equally confirmed opponent of (3), but there was no difficulty making
common cause with all the parties on interoperability. My own CogPrints
archive will now be rewritten according to the agreed upon Santa Fe
conventions in such a way as to turn it into generic archive software,
ready to be mounted by any university so all its researchers in each of
its departments can publicly archive all their papers.
The new archive software should be available for adoption for free for
all within six months. If a significant number of universities
worldwide mount and use it, the freeing of the research literature in a
global public archive could in principle take place before the end of
the year 2000. (The rest is only a question, having led them to the
waters of public archiving, of what will induce the research cavalry to
drink; the push from the institutional library serials budget crisis,
together with the pull from the prospect of universal barrier-free
access and impact just might do the trick, now that interoperability is
ensured.)
See:
Harnad, S. (1998) On-Line Journals and Financial Fire-Walls. Nature
395(6698): 127-128
http://www.cogsci.soton.ac.uk/~harnad/nature.html
Harnad, S. (1998) The invisible hand of peer review. Nature [online] (5
Nov. 1998)
http://helix.nature.com/webmatters/invisible.html
--------------------------------------------------------------------
Stevan Harnad harnad@cogsci.soton.ac.uk
Professor of Cognitive Science harnad@princeton.edu
Department of Electronics and phone: +44 23-80 592-582
Computer Science fax: +44 23-80 592-865
University of Southampton http://www.cogsci.soton.ac.uk/~harnad/
Highfield, Southampton http://www.princeton.edu/~harnad/
SO17 1BJ UNITED KINGDOM
NOTE: A complete archive of this ongoing discussion of "Freeing the
Refereed Journal Literature Through Online Self-Archiving" is available
at the American Scientist September Forum (98 & 99):
http://amsci-forum.amsci.org/archives/september98-forum.html
------------------------------------------------------------------
"Free the Journal Literature Through On-Line Self-Archiving"
See:
http://www.cogsci.soton.ac.uk/~harnad/Temp/bbc.htm
Universities cannot afford to buy the books they need because they
spend so much money on subscribing to research journals, most of which
they cannot afford to subscribe to either.
Books are an insoluble problem, because their authors, along with their
publishers, understandably wish to be paid for them. But the papers in
research journals are give-aways. Their authors don't seek or make a
penny from them. So why do their universities (who pay their salaries
to perform and publish their research) have to pay to get them back
again? Because publishing in paper costs money, and because quality
control costs money: The papers submitted to journals must first undergo
"peer review," sometimes several rounds of it, before they are ready to
be accepted, edited and then certified as published articles.
But the peers who referee the papers, do it for free too. So that does
not cost money. It is only the administration of the refereeing, plus
the editing, that costs some money; and of course printing and
distribution in paper.
A way is at hand now to free this give-away literature at last. Last
month in Santa Fe the "Open Archives" Initiative
<http://vole.lanl.gov/ups/ups.htm> was formed so as to agree on a way
for universities to archive all their research papers publicly in
"interoperable" online archives on the World Wide Web.
All archives complying with the Santa Fe Convention can then be drawn
together into a global "virtual library" that will contain the contents
of all existing research journals, free for all.
The key to this initiative is self-archiving: Authors will be able to
self-archive all their give-away papers in their own university open
archives -- both the pre-refereeing draft and the final, accepted
refereed draft -- and those archives will all be integrated seamlessly
into a global virtual archive.
What will become of the publisher's version? As long as there is a
demand for the paper version, that can continue to be sold; and if the
publisher produces an enhanced online version that people are willing to
pay for, that can be sold too. But the author's give-away version will
be available free in the open archives too, and it is very likely that
once users come to rely on that, as 50,000 physicists daily have
already come to rely on the 100,000 papers self-archived so far in the
Los Alamos Archive, journal subscriptions will begin to feel growing
cancellation pressure. As universities cancel their subscriptions, they
can redirect about 80% of that saved money toward buying the books they
need; 20% can be reserved for the day when journal publishers have
phased out paper and all its associated costs, and are no longer
providing a product (the paper), but are only performing a service
(quality control and certification). That annual 20% will be more than
enough to pay for that service for each university's own authors'
annual research papers.
Only three things could stop the freeing of the research journal
literature. One is if publishers try (and succeed) to prevent their
authors from self-archiving by forcing them to sign copyright transfer
tgreements that explicitly forbid it. My guess is that the this would
be so transparently in conflict with the best interests of research,
researchers, universities, and the society that benefits from the
fruits of the research, that it will not even be attempted. Indeed, the
American Physical Society, the publisher of the highest impact journals
in physics, have already written the right to self-archive into their
copyright agreements in the face of the collossal success of the Los
Alamos Archive.
The second possible obstacle is that universities fail to provide the
means for their authors to self-archive. At Southampton we will by
February make available (for free of course) Santa Fe compliant
software that will make it possible to create open archives at any
university immediately <http://cogprints.soton.ac.uk>. If the enhanced
impact that this will give their research and researchers is not enough
incentive to ensure that universities will create open archives, then
their journal budget crises surely will be.
The last obstacle is authors themselves: Once the copyright problem is
eliminated and the archives are ready to receive all their papers, will
authors take the trouble to do it? I think so, and let me count the
reasons: (1) The self-archiving will be made so easy that the effort
will be virtually zero; (2) given a tiny budget, even that minimal
effort could be made by proxy by computer-savvy students for professors
who are too busy or too old to learn new tricks; (3) 100,000 physicists
are already doing it, and their numbers are growing.
In principle, the journal literature could be freed before the end of
2000. If it is not, it will not be for lack of having led the horses to
the water.
--------------------------------------------------------------------
Stevan Harnad harnad@cogsci.soton.ac.uk
Professor of Cognitive Science harnad@princeton.edu
Department of Electronics and phone: +44 23-80 592-582
Computer Science fax: +44 23-80 592-865
University of Southampton http://www.cogsci.soton.ac.uk/~harnad/
Highfield, Southampton http://www.princeton.edu/~harnad/
SO17 1BJ UNITED KINGDOM
This archive was generated by hypermail 2b30 : Tue Feb 13 2001 - 16:23:05 GMT