Classification and tracking of hypermedia navigation patterns
Résumé
We consider the classification and tracking of user navigation patterns for closed world hypermedia. We first propose a series of features characterizing different aspects of the navigation behavior. We then develop Hidden Markov models and a variant of these models called Multi-stream Hidden Markov models to track on line the behavior of a user. We also provide experimental results for the recognition of pre-defined user behaviors, using a home made basis.