A system and method of predicting items likely to appeal to a user, based on data received from a remote site. One or more local systems are provided, along with a remote system, said local systems communicating with said remote system over a network connection. Each local system maintains a set of characterized items. A local processor predicts an appeal of an item to a user based on the comparison with the set of characterized items with, e.g., user preferences. The user may interact through graphical user interface. At least one user-preference predictive algorithm may be employed to present items predicted to appeal to the user.