Le groupage flou avec AFSA: Methodologie et application a l'analyse des sites web

This work discusses the application of the artificial fish swarm algorithm (AFSA) to fuzzy clustering. AFSA is used to optimize the performance of the fuzzy C-Means (FCM) algorithm and improve its clustering results. A modified AFSA with adaptive visual distance and adaptive step is proposed. An alternative fuzzy clustering method that does not require fixing the number of clusters using prier beliefs is also introduced in this work. This newly proposed automatic clustering method empowers the existing improved artificial fish swarm algorithm (IAFSA) by the simulated annealing algorithm (SA). Computer simulations are performed to compare this new method with a few state-of-the-art clustering algorithms by using several synthetic and real-life datasets. The experimental result proves that this newly proposed method performs better than other existing algorithms in terms of both accuracy and robustness.;Key words: Web auditing, Fuzzy clustering, Fuzzy C-means, Artificial fish swarm algorithm, Web access log data.;In this work, we have also applied the proposed method to audit web sites. A new measurement index for similarities between user sessions is presented. The proposed index combines both the user browsing time and website hierarchy structure. To validate our new approaches we have executed this newly developed method on the experiment data. The results are encouraging and show the potential of our fuzzy clustering approach to assist in auditing web sites.