Web Personalized Service Oriented User Modeling Technology

In the past decade, the information on Internet increased by degrees of index number. The netizens found it more and more difficult to get valuable information and materials that they want very quickly. The huge amount of Internet information resulted in the problem of information overloading and information-mazing, but what the Internet users need is to find the proper materials very soon, without wasting much time on searching. So, personalized service, which can solve this contradict to a certain extent, has now became a hot research field. In personalized service, the User Interests Modeling technology as the key issue is mainly to study how to mine the user's interests more accurately, analyze the user's interest themes, and construct the user's interesting models, according to the web pages or other data provided. In this paper, we research the User Interests Modeling technology mainly from the following aspects, and prove the validity of our research through experiments.①Mining the user's interests automatically and accuratelyIt's important to mine the user's interests automatically and accurately in modeling user's interests. In order to achieve this goal, an improved hybrid clustering algorithm(IHCA) is proposed in this paper based on the improving for the hybrid clustering algorithm(HCA). The improvements are mainly the following two aspects: on the one hand, to improve the automatically calculation of the number of clusters algorithm in order to increase the stability and accuracy of the algorithm, and using automatically calculation of the number of clusters algorithm to provide the initial cluster number for improved hybrid clustering algorithm; on the other hand, to improve the clustering part of HCA algorithm in order to not only increasing the accuracy of the HCA algorithm, but also adjusting the cluster number dynamically in the process of evolving, and to get the cluster number and cluster partitions more accurately at last.②Evaluating the accuracy and rationality of user's interest points mined for user's interestMining the user's interest points with the improved hybrid clustering algorithm is searching the reasonable cluster number and cluster partitions. So a new fitness function combined with the concept of clustering validity function is proposed in this paper for the fitness function of the improved hybrid clustering algorithm. this fitness function can evaluate the cluster number and cluster partitions more accuracy, implement the function of evaluating user's interest points.③Construct the user interest modelsThe final step in the user modeling system is build user interest models with the information mined. The user interest models are described by the two class vector model in this paper, and the process of build user interest models based on the user dictionary and two class vectors model is described in this paper in detail.④Experimental verificationFinally, the experiments are done, It's proved that: firstly, the improved hybrid clustering algorithm can find clustering a reasonable number of clusters more accurately; secondly, the improved hybrid clustering algorithm can get the cluster partitions more accurately; thirdly, fitness function of the improved hybrid clustering algorithm can be a more accurate evaluation of the number of clusters and cluster partitions; fourthly, user interest models can help the personalized search system to provide well services.The User Interests Modeling technology proposed in this paper can be mainly applied in quick and intelligent personalized search service. but the mainly technology can be expanded, so it also can be applied in personalized recommendation systems and intelligent user information modeling system in other fields, aided designers and managers to analyze the user information, if changing the fitness function and user interest models appropriately. It has both good academic value and good application value in many domains.