The relationship between toc, searches, abstracts, and full article
downloads (both html and pdf) is a very interesting and difficult one.
The factors seem to be very different for different journals. It seems
in some cases users only go to search or toc if they intend to download,
and for some quite the opposite. To what extent this is
journal-dependent or field-dependent requires further analysis. As an
example of the differences found, I quote some recent APS stats from my
institution:
PRB,condensed matter: tocs 13,834; searches 2501; Abs. 3529; arts.3794
PRE, statistical, etc.: tocs 2034; searches 1218; Abs. 1863; arts.2034
I do not think one can as a matter of course add them up for a
particular journal and get a meaningful total. The providers, and those
who run their own servers, do have the log data from which this
information could in principle be calculated, but I do not know if
anyone has. I can imagine it would be a very formidable task, but if
anyone would like to talk to me about it, I'm interested.
Arthur Smith wrote:
>
> David Goodman wrote:
> >
> > The use of arXiv during March 2002 was 2.5 million connections to the
> > main http://arXiv.org site, not counting mirror sites. That's a factor
> > of 100:1.
> > (By no means does everything in arXiv correspond in subject to Phys Rev
> > D, but that factor should be compensated for by the mirrors.)
>
> We've been over this before...
>
> Those are not comparable numbers. I was talking about only full article
> downloads, not searches, abstracts, tables of contents, etc, which give
> multiplier factors of 10 or more if you want to compare a total number
> of "connections". High energy physics is probably only about 1/5 of
> what's at arxiv.org these days - cond-mat and astro-ph are very big.
> Also hep-th, hep-ph and company let through about twice as many HEP
> articles as we publish in Phys Rev D. And I doubt the mirrors (except
> the original Los Alamos one) get more than 10% of the traffic unless
> arXiv.org is deliberately redirecting it. We run a mirror at
> aps.arXiv.org, and I know what the numbers are for it.
>
> When you put it all together, our Phys Rev D is maybe a factor of 2
> behind arXiv.org in downloads per "published" paper, and we're
> comparable to arXiv.org or have higher numbers for the other areas of
> physics. It's not anywhere near a factor of 100:1. And it's been
> reasonably stable this way for the last 3-4 years.
>
> Arthur
>
> PS [Sorry Marty, I couldn't resist - I'll be quiet now :-)]
--
David Goodman
Research Librarian and
Biological Science Bibliographer
Princeton University Library
Princeton, NJ 08544-0001
phone: 609-258-7785
fax: 609-258-2627
e-mail: dgoodman_at_princeton.edu
Received on Mon Apr 22 2002 - 23:24:23 BST