[open-bibliography] Bibsoup - can it handle large files?

Tue Jan 31 04:16:52 UTC 2012

Peter Murray-Rust <pm286 at cam.ac.uk> wrote:

> I have uploaded our Centre publications list to
> http://bibsoup.net/petermr/ucctest2. 

Good to see.
Its a bit disappointing that this dataset contains no links to text or even abstracts.
This was facilitated in the legacy bibserver by scripted links to Google Scholar, e.g.
this simple title link

http://scholar.google.com/scholar?hl=en&q=++Ligand-Target+Prediction+using+Winnow+and+Naive+Bayesian+Algorithms+and+the+Implications+of+Overall+Performance+Statistics

gets the user to an abstract of the first item

Nigsch, F and A Bender and Jenkins, JL and Mitchell, JBO (2008)
Ligand-Target Prediction using Winnow and Naive Bayesian Algorithms and the Implications of Overall Performance Statistics
vol. 48, 2313-2325

As a feature request, I suggest a simple display option in any BibServer display to show scripted links to Google Scholar or other desired service e.g.
Microsoft Academic or MathSciNet.

Automated provision of such links, and better still, making it easy for users (or machines) to follow these links and harvest 
metadata  including e.g. abstracts from the sites they point to, would add great value to typical BibServer displays.

This task could be facilitated by a BKN Linkserver, i.e. a BKN node dedicated to matching and enhancing biblio data uploaded anywhere to BibSoup.
This is a basic repetetive task, given a single bibitem, of whatever standard, to enhance it as best as possible by machine matching
with various sources, and thence create a multiply linked hybrid record  of superior quality.  I've been working towards this with experiments in Berkeley.
Would be glad to have assistance on this. Experience with federated searching or multithreading should be helpful to build a prototype. I think the main 
issue is handling varied responses from multiple sources and offering them to user is a way that is readily mashed up to provide the composite record.

--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