Ranking the Web Frontier
Source:
WWW2004, ACM Press, New York, NY, p.309 - 318 (2004)
URL:
http://mccurley.org/papers/1p309.pdf
Keywords:
Ranking, PageRank, Hypertext
Abstract:
The celebrated PageRank algorithm has proved to be a very effective
paradigm for ranking results of web search algorithms. In this
paper we refine this basic paradigm to take into account several
evolving prominent features of the web, and propose several algorithmic
innovations. First, we analyze features of the rapidly growing
“frontier” of the web, namely the part of the web that crawlers
are unable to cover for one reason or another. We analyze the effect
of these pages and find it to be significant. We suggest ways to
improve the quality of ranking by modeling the growing presence
of “link rot” on the web as more sites and pages fall out of maintenance.
Finally we suggest new methods of ranking that are motivated
by the hierarchical structure of the web, are more efficient
than PageRank, and may be more resistant to direct manipulation.
