[current]
[Gary Flake, Steve Lawrence, C. Lee Giles, Frans Coetzee; NEC Research Institute]
Link structure alone identifies communities
Solely by analyzing the link structure of the web (the pattern of hypertext links between pages), communities of highly related information can be identified. For example, communities related to different interests such as Pokemon or stem cell research.
Independent of textual context
Previously, researchers could match pages to an existing topic that they have defined by analyzing the text on the pages. This method is limited because the topic must be known in advance, it is very difficult to retrieve all pages in a timely manner, and because the method depends on the specific words used (e.g., pages in a different language or using different words for the same concept may not be identified).
[pointer via
chris and I will not point to the Spiegel article which
Eberhard says is a more-or-less translation of this nature acrticle (
Search engine makes social calls)]
This also links nicely with
VisIT which I linked some time ago. A graphicaöl analysis of search-results, and most importantly their inter-linking structure.
[ by Martin>]
[]
[]
similar entries (vs):
- some links on lists of links via Eberhard (# 9%)
- VisIT is a revolutionary new way to search the Internet. (# 8%)
similar entries (cg):
Martin Spernau
Big things to come (TM) 30th Dez 2002
