The goal of event tracking is that of discovering and following postings, discussions, or more generally, tangible expressions of interest about an event in a temporal text stream. We propose a scalable framework for tracking a set of user-specified events.
This paper describes our novel approach to the quanitification of the level of topic propogation in the blogosphere. It tries to answer one key question: How many people should know about a subject before it becomes prevalent? Our model uses graph-theoretic representations of link structures, allowing us to deduce a "percolation threshold."