Open Peer-Review as Web's Self-Organization Force

Department of Computer Science,
SUNY at Stony Brook,
Stony Brook, NY 11794, U.S.A.
February 2000

maxim@cs.sunysb.edu

Abstract

We propose a novel system to support an ultimately open and scalable distributed collaborative Web categorization and ranking scheme that follows the quality ensuring peer-review spirit of academia and the Open Source movement. It is a categorization and ranking metadata standard with a unique efficient rank computation model that enables cooperative Web directory construction with automatic and democratic generation of a dynamic hierarchy of accountable experts. Our method both extends and combines in a coherent framework the individual powers of essentially all existing approaches to solve the large-scale Web information retrieval problem including directories and search engines with global linkage analysis. The proposed approach stands on a solid mathematical foundation subsuming the underlying theory of the Google search engine [17]. An implementation of our method can utilize simultaneously both the data currently available on the Web and the proposed metadata, while from the start delivering higher quality Web searching services than state-of-the-art search engines and directories. Thus it provides a smooth migration path to move the Web towards a much more organized and valuable information service. This paper presents the architecture and major algorithms of our system and compares it with existing search engines and directories.