When one of SwiftFile's predictions is correct, the effort required to file a message is reduced to a single button click. An incremental classifier is used to allow SwiftFile to continuously update its model of the user.
IBM have performed a number of static and dynamic experiments with SwiftFile as well as some initial user studies. The results show that the shortcut buttons provided by SwiftFile are useful between 80% to 90% of the time, and this performance level is maintained over time despite the constantly-changing classification problem. We also find that the bootstrap period for the classifier is extremely short and that SwiftFile is useful from the moment it is first installed.
Comments