jump to navigation

DBLP Computer Science Bibliography January 6, 2009

Posted by JR Dixey in links.
Tags: , ,
comments closed

If you’re researching anything to do with HCI or any other aspect of information communication technologies, this site should be one of your stops. http://www.informatik.uni-trier.de/~ley/db/index.html

Funded by Microsoft Research and something called the Very Large Database Foundation (who knew?), and hosted at the University of Trier, this site indexes more than 1 million articles and conference proceedings authored by more than 12,000 individuals, cross-referencing them based on co-author, year, and venue (journal where published or event).

There is a faceted search option, for Ranganathan fans (http://dblp.l3s.de/?q=&newQuery=yes&resTableName=query_resultT0FSWN), and if you’re looking for fast, I mean very fast, results, the CompleteSearch interface will “zoom in” on all the articles indexed in the database by a particular author as you type the author’s name. (http://dblp.mpi-inf.mpg.de/dblp-mirror/index.php).

Most astonishing of all is that even though this is, ostensibly, a database, there is no there there – to quote Gertrude Stein – in that the data is not aggregated into a single database file but distributed throughout thousands of text files, with the core content — journal TOCs — hand-entered by human beings.

Yes, indeed. Every journal’s Table of Contents is hand-typed into a predetermined HTML format that can be easily parsed by programs written in C, Perl, and Java, “glued together by shell scripts” (to quote from the detailed FAQ). Just goes to show all us 2.0 Web babies that sometimes, the old fashioned approach can work very, very well.

Automated Accuracy Rating for Wikipedia Entries September 7, 2007

Posted by JR Dixey in news.
Tags: , ,
comments closed

Interesting project out of University of California at Santa Cruz: software that can (purportedly) track the accuracy of information inside Wikipedia entries. The software, developed by Luca de Alfaro, a Stanford Ph.D. who is currently an Associate Professor of Computer Science at UCSC, monitors the trustworthiness of Wikipedia entries based on the reputations of their contributors, and modifies the display of Wikipedia articles based on realtime, calculated contributor metadata.

Quoting from Professor de Alfaro’s own statement about the project, it computes the “trust value of each word of a Wikipedia article” based on the reputation of its original author and the reputation of all its subsequent authors. If you’re curious to try it out, there’s a description and demo of the system available at the UC Santa Cruz web site, using a dump from Wikipedia created in February of 2007.