Re: New Ranking of Central and Institutional Repositories

From: Stevan Harnad <harnad_at_ecs.soton.ac.uk>
Date: Thu, 14 Feb 2008 08:26:59 -0500

Begin forwarded message:

      [Stevan, please post this for me. Thanks very much.]

      Hi Arthur, Mark,

      We comment partly on the basis of what exists now, rather
      than what
      could be in place, I think. A colleague of mine has
      written a script
      that automatically includes author's bibliographies in
      their personal or
      departmental web pages, just by using a link in a PHP (or
      other) server
      include in an HTML tag. This saves them lots of work and
      encourages
      deposit.

      Let's suppose that every academic did this (as I suspect
      they don't,
      even if able). Could we anticipate that academics might
      want to look at
      each other's research profiles much more, given that
      they'd have a
      direct link to the output? I wonder, do you have any
      statistical
      information on this, Arthur, to say whether this actually
      works? If it
      became more widespread, as a general culture rather than
      just in the odd
      institution, would these links gain value that they don't
      presently
      have? I'm not arguing with you in principle, Arthur, as I
      have neither
      evidence nor motive, but I'm interested to know what you
      think: how can
      we know the future value of links if at present their
      application is
      haphazard? I'm sure that the haphazard nature of any
      "service" puts
      people immediately off using the occasional parts of it
      that do work,
      because these are difficult to predict. Just speculations
      on my part
      perhaps, as the future is a bit tough to research!

      It's my hope that *any* ways to increase access research
      content, be
      that search engines, links or any other method, will grow
      as open access
      grows, so what seems trivial at present may one day be a
      useful way to
      promote content. Things change so fast that we do need to
      entertain this
      sort of speculation, I think. For the moment, of course,
      our stats here
      for example do seem to show, as Arthur says, that search
      engines (well,
      just Google) are way ahead of everything else.


      Talat

      -----Original Message-----
      From: American Scientist Open Access Forum
      [mailto:AMERICAN-SCIENTIST-OPEN-ACCESS-FORUM --
      LISTSERVER.SIGMAXI.ORG] On
      Behalf Of Arthur Sale
      Sent: 13 February 2008 20:47
      To: AMERICAN-SCIENTIST-OPEN-ACCESS-FORUM --
      LISTSERVER.SIGMAXI.ORG
      Subject: Re: New Ranking of Central and Institutional
      Repositories

      Mark Doyle

      The dominant institutional repository softwares ALL have
      such simple
      identifiers. The softwares are not badly designed.

      But that does not obscure the essential point that inward
      bound links on
      the
      open web are a minor contributor to a repository success.
      Many links
      that
      are used are not exposed on the Web, and are simply kept
      in an EndNote
      file,
      or bookmarked. People use institutional repositories via
      search engines,
      not
      links.

      Subject repositories are subject to different rules.

      Arthur

            -----Original Message-----

            From: American Scientist Open Access Forum

            [mailto:AMERICAN-SCIENTIST-OPEN-ACCESS-FORUM
            -- LISTSERVER.SIGMAX

            I.ORG] On Behalf Of Mark Doyle

            Sent: Thursday, 14 February 2008 6:41 AM

            To: AMERICAN-SCIENTIST-OPEN-ACCESS-FORUM --
            LISTSERVER.SIGMAXI.ORG

            Subject: Re:
            [AMERICAN-SCIENTIST-OPEN-ACCESS-FORUM] New

            Ranking of Central and Institutional
            Repositories


            Hi,


            On Feb 12, 2008, at 4:38 PM, Arthur Sale
            wrote:


                  This brings me to the second
                  point: Repositories were not set
                  up to

                  provide linkage, and if they were
                  to be in the deep web apart from

                  being harvestable, their utility
                  would be only slightly weakened.


                  Indeed this is exactly the
                  situation with most of the PhD
                  thesis

                  repositories in Australia. The
                  federated site is open to

            the Web, and

                  a very few thesis sites like my
                  university's, but most university

                  repositories are simply in the
                  deep web, accessed only by the

                  federated harvester. This is the
                  Australasian Digital

            Theses Program,

                  also listed in the Webometrics
                  top 200. I haven't heard 30+

                  universities complaining about
                  the loss of links.


            I think this is poor design.  Depending on
            bookmarked URLs

            and 'browse by name' is a rather fragile
            infrastructure. One

            of the reasons that a central repository like
            arXiv.org is so

            successful is because Ginsparg, in his
            wisdom, came up with

            short, somewhat meaningful identifiers (new
            arXiv ids are now

            slightly less than ideal, but at least you
            can tell right

            away when something was first entered into
            the repository)

            AND provided for the ubiquitous linking to
            arXiv.org via the

            /abs/ID URL. These URLs have been stable
            since they were

            introduced in 1994 when the web interface was
            introduced

            (even after the xxx.lanl.gov -> arXiv.org
            transition). This

            has allowed the arXiv staff to insert
            clickable links into

            the PDFs and people to trivially link to the
            arXiv version of

            a work. On the publisher side, considerable
            effort has been

            put into introducing DOIs which again make it
            easy to provide

            interlinking between scholarly articles. Some
            publishers

            (like APS) have easy to create DOIs from the
            usual (journal,

            volume, page) metadata (or even URLs that
            don't depend on a

            DOI), while others are more opaque. CrossRef
            levels the

            playing field though and makes DOIs easily
            discoverable. In

            any case, one should not underestimate the
            usefulness of

            having simple identifiers that map
            algorithmically to permanent URLs.


            Best,

            Mark


            Mark Doyle

            Assistant Director, Journal Information
            Systems The American

            Physical Society
Received on Thu Feb 14 2008 - 13:30:18 GMT

This archive was generated by hypermail 2.3.0 : Fri Dec 10 2010 - 19:49:13 GMT