| Internets provide massive information resources for us, which is gradually becoming an indispensable part in our life. Personalized service will be the development trend in the future that the information will serve. It is the foundation to obtain the customer personalized interest and the information need in carrying out personalized service. The technology of user modeling is placed on the core position in all the other related technology in personalized service. Starting from the viewpoint of user modeling based on the humanized website interface design, the main research content in this paper is described as follows.1. BP neural networks and its improved algorithmIn view of the low convergence rate, the difficulty in obtaining optimal solution in the overall situation because of the local minimum, and the other unfavorable factors such as being more sensitive to the initial value selection of the BP Neural Network algorithm, some researchers put a variety of measures to improve algorithm. Among them are join the momentum method, conjugate gradient method, variable step method and Levenberg-Marquardt BP algorithm which are typical methods. On the basis of the those previous research work, this paper proposed an improved algorithm of the VLBP called as IVLBP by introducing two thresholds of error square increment and decrement to have a fast convergence rate. At last, the comparative simulation results demonstrate the effectiveness of the IVLBP algorithm2. User modeling of personalized web interface based on BP neural networkTo realize the humanized website interface style, we study the relationship between user factors and the web interface factors by investigation, and select some key factors from the complicated user and web factors. By use of BP neural networks, we have established a user model based on those key factors which can be used in the design of the humanized web interface,to enhance the user's satisfaction for the style and arrangement of the website interface. |
