Personalized Information Services, User Interest In Migration Research

With the development of information technology people have to live in the"information explosion"age. Today, what we are facing is no longer lack of information and difficulties in use, but how to make full use of the exponential growth of data. So data mining techniques appears and have emerged to become a hotspot today. Web data mining now is an important part of data mining. It is a way to scan the information which is useful, potential and not easy to be found from various Web data with large amount.Due to the continuous development and spread of the network, there has been a trouble that the user can not deal with and make full use of this information with large amount in time. It is called"information overload". Too much information troubles the users that they can neither show their needs for the information resource, nor find out these resources efficiently. That is called"lost in information". Users can only be a passive receiver of information instead of seeking the very information they are interested in. What we want can not easy be found meanwhile what we do not want rushes in. For this, the users apply for getting the information of interest and meeting the requirement the demand. For information service providers it is also a new opportunity, in order to adapt to market competition, and create new competitive advantages and profitability of space, we should be able to develop a new way to meet the user's interests and hobbies and personalized information service. Due to the need between both sides, personalized information services appear. The key to personalized information services are also based on the collection of users'interests, then improve website design and structure according to these information. And to provide personalized and targeted services, such as information on custom, advertising recommended etc. For this, I present a new mode for collection of user interest, which is a model based on feedback from users interested in implicit and explicit.The user's interest is not static, as time goes by, some interests in the original will be out of date and some new interests will appear. So how to capture the complete set of new users interested is becoming a key. This paper aims at the changes of user's interests and how to know their new interests. Gradual forgetting and timetable window are the two common ways we usually use. This passage will introduce a new way of mixed-model. According to the research in physiology we sort the interests into two different kinds ------ long-term interest and short-term interest. We apply sliding window algorithm to short-term interest and gradual forgetting to long-term interest. It feeds back the change of user's interest. At last, a personalized information service system is introduced. This system can not only trace the change of the user's interest but also predict the interest of the users.