Re: Self-Archiving Incentives: Download Impact Counts

From: Arthur Sale <ahjs_at_ozemail.com.au>
Date: Thu, 25 Nov 2004 00:03:17 +0000

We have been busy developing user-friendly statistics to encourage our
contributors by giving them something (information) back, thereby tackling
THE problem of OAI repositories. Since I believe in working off the access
logs (there is so much information already there) rather than adding
counters to Eprints, we asked for access to Melbourne Uni's source code for
their stats package (thanks unimelb), and set about modifying it
substantially to our needs. We also added the open-source package awstats
(Advanced Web Statistics) for our own site management purposes, only
slightly tweaked.

Take a look at the result of Stage 3 of this development, either by going to
any document in our archive and clicking the Detailed Statistics link at the
bottom (the normal route for contributors and searchers I suspect) or via
the link (Detailed statistics) on the entry page to the archive
http://eprints.comp.utas.edu.au/ which may be used by potential
contributors. You can find the link to awstats on the entry page too.

We have been client-oriented throughout, with an emphasis on contributors
rather than searchers:

* Made the whole sub-site oriented towards document downloads as the
main measure of success, while displaying information about metadata page
views for interest.
* Made all the histograms pretty and colour-coded, plus our own set of
nice flags in an 18x14 pixel format for a country-of-origin analysis.
Pleasant to look at.
* Added UTas Intranet to the IP address table and the flag directory
as a pseudo-country and removed the addresses from Australia. Why? Not
because Tasmania is an island! Intranet enquiries are not what we are really
for and they distort the picture of external Australian accesses. Authors
and their students often look at their documents, and they are used
internally in demos to promote eprints and train contributors and editors.
* The history of views for a document is continuous from the latest to
the first recorded access, even if months with no views intervene. You can't
see it yet on our archive (only two months of stats), but the code is there.

* Changed order of items, revised text, added explanatory text,
generally made navigation self-explanatory, etc.

I'll now take a breather, before starting on designing author-oriented
statistics, to allow authors to see a summary of all their paper downloads,
and maybe their rank in the author list. Whatever. I haven't done the needs
analysis yet. Since all this stuff so far works off the logs, it should all
be reasonably portable across eprint software and repositories.

Arthur Sale
Professor of Computing (Research), University of Tasmania
127 Tranmere Road, Howrah, Tasmania 7018, AUSTRALIA
Phone (03) 6247 1331 (International replace '(03)' by '+61-3-') or Mobile 04
1947 1331
Received on Thu Nov 25 2004 - 00:03:17 GMT

This archive was generated by hypermail 2.3.0 : Fri Dec 10 2010 - 19:47:42 GMT