Automatic Categorization of Users Based on their Web Navigation

Web navigation can be researched in several ways. Apparently the most popular is the evaluation of navigation using user experiments. Specialized peripheral devices such as an eye-tracker allow us to get a lot of information in real time about the user. There are also papers about personalizing the web and predicting user properties by observable variables from navigation. Personalization can be helped by knowing the cognitive style of the user, that is, how he thinks and organizes information, which can in turn make him more efficient when navigating the web. However, there are several separations of cognitive styles that focus on different human aspects, such as his or her orientation in space or the way he or she observes situation and organizes information.

In this work, we are predicting the cognitive style of a user based on metrics evaluated from our experiment. Using classification method we can evaluate the cognitive style according to the chosen Wholist-Analyst and Verbal-Imager dimensions based on user’s navigation behavior. Further, we are suggesting positive changes for effective navigation on the web based on individual cognitive styles determined from acquired user data during our experiment.