Drag-and-Guess: Drag-and-Drop with Prediction

Drag-and-guess is an extension of drag-and-drop that uses predictions which is based on application specific knowledge. As the user begins to drag an object, the system predicts the drop target and presents the result to the user. When the target is hidden in a closed folder or beneath other windows, the system makes it temporarily visible. This frees users from manual preparation such as expanding a folder tree or uncovering the target location. The user can accept the prediction by throwing the object, which then flies to the target. Or, if the prediction is unsatisfactory, the user can ignore it and perform the operation as usual. We built three prototype applications (email client, spreadsheet and overlapping windows) to show that DnG is useful in many applications. Results of the user study show that the proposed technique can improve task performance when the task is difficult to complete manually and reasonable prediction algorithm is available.