Knowledge acquisition using neural networks for intelligent interface design

Several methods for using neural net models for intelligent interface design are investigated. All of the methods use a bottom-up approach in which the intelligence incorporated into the interface is based upon the inputs from individual users. In the first piece of research, the input came from the mental and external operators, closely coupled to the keystrokes, and these were mapped to high-level cognitive goals. In the second piece of research, the input came from individual commands and these were mapped to the efficiency of the command sequences. In the last piece of research, information from bulletin board messages could be filtered by using the dictionary of words from the messages and outputting the subcategory. This experimentation demonstrates some of the possibilities of using neural nets in intelligent interfaces: the neural net can extract information from the user, such as cognitive goals, strategies, command sequences, and information needs, which can be used to provide intelligent advice to the user or to filter intelligently the incoming information