User Interest Learning in Pervasive Computing Environment

The advent of pervasive computing puts forward a new challenge for individual information research. With the explosion of information on the Internet, finding information relevant to a user´s interest can be a time-consuming and tedious task. User interest learning plays an important role in information personalization. In this paper, a learning approach to acquire and update user interest is proposed. The approach firstly models user profile as feature vectors. Then pervasive device captures user´s implicit feedback based on his/her reading behavior and delivers it to the server. At last, the server infers user interest and updates user profile by adjusting the weights of features to keep track of the dynamic change of user interest. The experiment suggests that the way of implicit feedback in the approach is effective and the precision of the information given to users is encouraging.