An on-line web visualization system with filtering and clustering graph layout

Abstract

A Web graph refers to the graph that is used to represent relationships between Web pages in cyberspace, where a node represents a URL and an edge indicates a link between two URLs. A Web graph is a very huge graph as growing with cyberspace. To use it for Web navigation, only a small part of the Web graph is displayed each time according to a user's navigation focus. The graph layout has always been a challenge for visualizing systems. In this paper, we present a visualization system of an online Web graph, together with the methods for clustering and filtering large graphs. In this system, a Web crawler process is used to get on-line information of the Web graph. Filtering and clustering processes reduce the graph complexities on visualization. In particular, the filtering removes those unimportant nodes while the clustering groups a set of highly connected nodes and edges into an abstract node. The visualization process incorporates graph drawing algorithms, layout adjustment methods, as well as filtering and clustering methods in order to decide which part of the Web graph should be displayed and how to display it based on the user's focus in navigation.