The Study Of Human-computer Interaction, User Modeling Methods

User modeling is an important study field of Human-Computer Interaction, it makes computer understand the way user apperceiving, the course user knowing and the mode user behaving through changing user knowledge model to computing model. On the basis of understanding user model, computer can change its own behavior mode to form the compatible interaction and interface and make cost least. So, modeling a better user model is the first step of implementing the natural interaction between system and user.User model has been widely used in related fields of Human-Computer Interaction, such as Adaptive Human Computer Interface, Intelligent Agent, Intelligent Tutoring Systems and Information Seeking. There are many different user modeling approaches and have made considerable research results.The theory of rough set appearing recently is the new theory that analyzes data. Its main idea is to deduce the decision rules or classification rules by knowledge reduction without changing the ability of classification. It has popularized the classical set theories in handling the problems that are uncertain or imprecise and can be used to describe such knowledge. Due to the uncertainty or imprecision in user model, we can introduce the theory of rough set to user modeling.The main contributions of this paper are:1 .Many user modeling approaches in last few years both at home and abroad are presented and the comparison between them are given. User modeling approaches can be divided into two sorts: One sort includes static user modeling and dynamic user modeling according to the way user model updating. The other sort includes Stereotype approach, Machine Learning approach, Bayesian Network approach, Neural Network approach, Logic-based approach and Fuzzy Set approach according to the techniques for constructing user model.2.The theory of rough set is introduced to user modeling, and the new modeling approach based on rough set is discussed in detail. It can describe the uncertainty or imprecision in user modeling, and it doesn't need any priori information beyond the data set that to be dealt with. In addition, it has good permeability and can be combined with other user modeling approaches.3.On the basis of the approach put forward, an Adaptive Car-Chosen Decision System (ACCDS) is designed. To realize the system, first, we use rough set theory to construct user model and get the user type decision rules, then we write these rules into the system. The system will change its stereotype to adapt to the user when it finds user type changing. Results of the experiment show that our rough set approach to user modeling of the system can meet the different user types thus can help the user to make decisions easily.4.A new method to classify the system users is presented. Using this method, we can get NSND, NSED, ESND and ESED four system user types. The new method thinks about both users' familiarities with computers and familiarities with the decision methods. Experiment results also verify its validity.