Demo: the Wikipedia, colored according to text trust

2 views
Skip to first unread message

Luca de Alfaro

unread,
Jul 30, 2007, 6:05:42 PM7/30/07
to ucsc...@googlegroups.com
Dear All,

we have set up a demo, where the text of Wikipedia articles is colored according to a computed value of "trust".
The trust of each word of a Wikipedia article is computed on the basis of the reputation of its author, as well as the reputation of all the Wikipedia users who subsequently revise the article.

For more information, and for the demo, see http://trust.cse.ucsc.edu/

Enjoy - and let us know if you have comments!

Best,

Luca

Dimitris Achlioptas

unread,
Jul 30, 2007, 6:57:38 PM7/30/07
to ucsc...@googlegroups.com
Hi Luca,

this is great! If you ever run into scalability issues with the
algorithms, let me know. I'd love to contribute. The paper I mentioned
some time ago is

http://citeseer.ist.psu.edu/mcsherry05uniform.html

Cheers
dimitris

Andrew Trapani

unread,
Jul 30, 2007, 7:56:46 PM7/30/07
to ucsc...@googlegroups.com
http://enwiki-trust.cse.ucsc.edu/index.php/Convention_of_Kanagawa

On OSX Firefox 2.0.0.5 I had trouble reading most of the links because
they are red text on orange or orange/red backgrounds. It is just
difficult to skim through low contrast text and background.

Luca de Alfaro

unread,
Jul 31, 2007, 12:14:17 AM7/31/07
to ucsc...@googlegroups.com
Dimitris,

we haven't implemented any page-rank like algorithm.... if we plan to, I will be sure to ask your advice.  In fact, we should chat in any case!  I am at UCSC tomorrow; most of the days otherwise I am on the Mountain View side of the bay (I am also leaving Wed to Tue for Wikimania).
Also tell me: are you an expert of approximate string matching?  Like the one used in bio algorithms for DNA sequences?  We currently use another method, but I think approximate string matching, followed by a detailed resolution of the differences, would give us a far more robust method of tracking text propagation...

Luca
Reply all
Reply to author
Forward
0 new messages