Extraction of dynamic user behaviors from web logs

Our paper proposes an approach which makes possible prediction of future states to be visited in k steps corresponding to k web pages hyper-linked, based on both content and traversed paths. To make this prediction possible, three concepts have been highlighted. The first one represents user exploration sessions by Markov models. The second one avoids the problem of Markov model high-dimensionality and sparsely by clustering web documents, based on their content, before applying Markov analysis. The third one extracts the most representative user behaviors (represented by Markov models) by considering a clustering method. The original application of the approach concerns the exploitation of multimedia archives in the perspective of the Copyright Deposit that preserves French's WWW documents. The approach may be the exploitation tool for any web site.