On Building Maps of Web Pages with a Cellular Automaton

We present in this paper a clustering algorithm which is based on a cellular automaton and which aims at displaying a map of web pages. We describe the main principles of methods that build such maps, and the main principles of cellular automata. We show how these principles can be applied to the problem of web pages clustering: the cells, which are organized in a 2D grid, can be either empty or may contain a page. The local transition function of cells favors the creation of groups of similar states (web pages) in neighbouring cells. We then present the visual results obtained with our method on standard data as well as on sets of documents. These documents are thus organized into a visual map which eases the browsing of these pages.