Automated user modeling for intelligent interface

The analysis of the user behavior is one important function of the intelligent user interface because, by analyzing the user behavior, it becomes possible to understand the user intention and release the user from tedious tasks which are often required to use a fast but low-level interface. The acquisition of the user behavior model is crucial. Most studies meant to realize an intelligent interface system only analyze superficial user behaviors, from which to automate the repetitions. Their user models tend to be simple and do not reproduce the behavior well enough. This paper presents a new framework that analyzes the computational processes activated by the user commands to build the user behavior model. An important feature of the proposed framework is the analysis of data dependency between the user commands. A user-adaptive interface system, CHpBoard, was developed to show the adequacy of this framework. It analyzes the I/O relationship between applications in the past task history, selects the next application, and creates scripts which enable complex task execution by a single command.