GWUM : une généralisation des pages Web guidée par les usages

The usage analysis of a Web site based on the extracted sequential patterns is often limited by the low support of these patterns. That is mainly due to the great diversity of the pages and behaviors. However, it is possible to group the majority of these pages in various categories during a preprocessing. Then, using these categories, rather than the URL, will allow us to discover "generic" behaviors. This article presents a methodology for Web usage mining that uses such a generalization of the URL. This generalization is based on a categorization of the URL using the information extracted from the Web users’ accesses to these pages. Then, we present an experiment which shows how the support of the extracted sequential patterns changes according to whether the patterns are obtained with or without this generalization. MOTS-CLÉS : Fouille des usage du Web, motifs séquentiels, classification