Back
to indexKrishna Bharat, Andrei Broder, Monika Henzinger, Puneet Kumar and Suresh Venkatasubramanian
Summary
The Connectivity Server provides the Google search engine with immediate access to the neighborhood of pages around a given page. The idea behind this approach is that pages that are strongly connected within a web subgraph are more likely to be related than weakly connected pages. The Connectivity Server provides this information quickly upon a query event, which is then used to rank the web pages efficiently. The linkage information is provided via a graph structure, which is stored as pointers to urls in a database. Each URL has a simple list of 'inlist' and 'outlist' links, which are links to and from a page, respectively.
This server provides the solid foundation upon which Google is built, which validates the efficiency and effectiveness of the approach. However, there is little room to build a crawler application from this material.
Keywords
connectivity, web links, web graph, web visualization, search
Methods
The connectivity server represents the web as a connected graph, and for every page (node), a list of 'inlist' and 'outlist' links are stored. The connectivity server assists with the connectivity analysis for queries generating a hub and an authority score for each page (normalized sum of the links from a page and to a page, respectively). The connectivity server is also able to generate an actual graph of the connections between different pages, real time.
Rating
6
Bibtex Entry
@proceedings{ bharat98,
author = "Krishna Bharat and Andrei Broder and Monika Henzinger and Puneet Kumar and Suresh Venkatasubramanian",
title = "The Connectivity Server: fast access to linkage information on the web",
text = "WWW7 Conference, 1998.",
year = "2000",
url = "http://www7.scu.edu.au/programme/fullpapers/1938/com1938.htm"
}