Combating Web Spam with TrustRank

Gyongyi, Garcia-Molina and Pederson

[vitaplease]

Combating Web Spam with TrustRank [dbpubs.stanford.edu]

The authors propose a technique to semi-automatically seperate reputable, good pages from spam. They select a small set of seed pages and then use the link structure of the web to discover other pages that are likely to be good.

Zoltan Gyongyi [www-db.stanford.edu] (Stanford University)
Hector Garcia-Molina [www-db.stanford.edu] (Stanford University)
Jan Pedersen (Yahoo! Inc.)

Interesting comparisons between Pagerank and Trustrank, where they claim Trustrank differentiates better between "good" and "bad" sites.

Interesting that one of the authors is from Yahoo, also given that we now also have Yahoo's Webrank [webmasterworld.com].

They also discuss what they call Inverse Pagerank
(giving importance to a page's outlinks, not its inlinks)

Trustrank also seems to be a term used by Shanti Braford discussed in this interview [kiruba.com] relating to Popdex's new features.

[edited by: vitaplease at 9:40 am (utc) on Mar. 23, 2004]

[Macro]

I like this idea of trustrank. I like it very much.

small set of seed pages

Therein may lie the crunch. What if they started off with amazon being a high trust site? Or kelkoo? (Don't even get into the Microsoft question! :))

What if a big "trustworthy" news site ran an expose on a dubious organisation and linked to their website? Would that site suddenly "acquire" trust?

However, I do look forward to seeing this trust idea develop. If a proper implementation can be rolled out it will be a big improvement on the Google PR that so many manipulate.