Web User Access Pattern Mining Based on Kohonen Neural Network

This paper provides a method for Web user access pattern mining based on kohonen neural network. User session vectors were first input to kohonen network, after training we found several clusters, compute the median of each cluster and characterize what the cluster represents with the URLs. When the online user requests URLs, a matching category is found according to the pages the user has accessed. Pages that the use has not accessed so far and will access are included as suggestions and links in the html to the user. This method is efficient in user access pattern mining and from it we can provide personalize services in order to succeed in the competition of Web services