Web Sessions Clustering with Artificial Ants Colonies

Nicolas Labroche, Nicolas Monmarché, Gilles Venturini

Abstract: In this paper, we apply AntClust, an ant based clustering algorithm, to the Web usage-mining problem. We define a Web session as a weighted multi-modal vector and we propose an adapted similarity measure between two sessions. We show that our method finds non-noisy clusters that help the interpretation of the results when it is applied on real Web sessions coded as "hits by page" vectors.