Generating Web Graphs with Embedded Communities

Vivek Tawde, Tim Oates, Eric J. Glover

Abstract: We propose a model to generate a web graph with embedded communities. The web graph will also have the properties that are observed on the web like bow-tie structure, node reachability and link distributions. We use a community as a basic building block to model the link distributions. We interconnect these communities using a model for interconnection that uses a citation matrix. The generated graph consisting of a combination of interlinked communities, is converted into a larger web graph by modifying the link distribution to achieve the desired distribution of the whole web. Using communities as the basic building block helps in maintaining local properties like resiliency and reachability of the simulated web graph.