Nauzer Kalyaniwalla

Title: Extracting Communities from Networked Information Spaces.

Abstract:
It has long been known that networked information spaces,
like the Web, contain many thematically unified communities.
In the case of the Web, these are groups of web-pages, on
similar topics, that are tightly linked to each other. Previous
work on identifying and extracting such coherent subsets on the
web have mostly relied on text-based analysis. We present preliminary
results for a new method to extract communities that is based purely on
link analysis. Our method relies on increasing the density within
communities and then extracting the community core with a clique-
finding algorithm. The experiments are performed on a synthetic
test-bed created, by us, specifically for evaluating graph-based
web algorithms.

This is work done jointly with Jeannette Janssen and Steven Gao.