[openbiblio-dev] some dogfood

Wed Nov 9 14:52:52 UTC 2011

http://bibsoup.net/pitman/bibserver_use_cases

reads from a JSON cache at

http://bibserver.berkeley.edu/tmp/bibserver_use_cases.json

of  Rufus's mediawiki creation

http://wiki.okfn.org/Projects/BibServer 

As a little dogfood exercise, let us demonstrate end-to-end capability for this use case (which should be added to the wiki!)
with live refreshes from the wiki. This should be simple to set up and maintain and should help get us past two blocker 
items:

1) provide way for authorized user to specify templates to customize BibServer displays for such special applications

2) provide way for authorized user to install scrapers for such special applications/sources using https://scraperwiki.com/

Mark: towards 2), please could you post a suitable variant of the 59 lines of python code at /accounts/projects/jpopen/wiki/0getwiki.py
to scraperwiki for all to see and streamline if possible for similar data piping from MediaWiki to BibServer?

Rufus, a couple of questions for you:
-- how exactly did you map the user case data from the Google Spreadsheet to the mediawikipage? 
-- what permissions or technical capabilities do I need to do this myself, i.e. to script another BibServer mediawiki page from a csv or similar file?

And a general UI question for the group:
--- how long of a list do you think it would be efficient to manage in mediawiki like this?  I think the answer is about 30 records, and that
 anything much bigger than this should be broken out hierarchichally.

This wiki2bibserver might provide a better setup for collaborative editing and display of the BibJSON spec and
other datasets than any we have found to date. Enabling bibliographies to be simply and efficiently
maintained on mediawiki could enlarge the community of users by engaging some of the growing number of mediawiki sites
used to manage academic collections. None of these I have seen have decent biblio display capability. See e.g.
http://wiki.math.toronto.edu/DispersiveWiki/index.php?title=Category:Bibliography&from=A for a typical source, most likely cooperative.
Generally, I think it is important for us to be connecting directly to currently deployed data management tools like wikimedia and Drupal
rather than trying to develop our own collaborative editing and CMS tools.

More for Mark:

Some simple UI improvements for the use case doc  would be to allow faceting by Tag and Status instead of the present empty facets.
Also, as part of 1), user should be able to modify the text "External links" which is not right in this case as some of the
links are internal. This is typical of how subject or keyword tags may be supplied by user in links. Because of this capability, which I really
appreciate, it may not be necessary to provide special support for subject or keyword tags in typical biblio data. Just advise users to stuff them
into links, which is where they belong. More advanced display would show a header record at the top of list for any subject or tag for which one
was provided in the dataset. The same trick can be used for tagging of other entities too.
It is very useful to retain the primitive BibServer capability of displaying just about any list of key-value records, with capability for
search and faceted browse.  I have any number of other not-strictly-biblio datasets I'd like to throw up like this on a BibServer, and others may like
to use it so flexibly too.

--Jim

----------------------------------------------
Jim Pitman
Professor of Statistics and Mathematics
University of California
367 Evans Hall # 3860
Berkeley, CA 94720-3860

ph: 510-642-9970  fax: 510-642-7892
e-mail: pitman at stat.berkeley.edu
URL: http://www.stat.berkeley.edu/users/pitman