The Research Of Personalized Chinese Information Service System For Digital Libraries

Internet has extended the scope of people's obtaining information. Meanwhile, WWW has organized all kinds of information into a huge information space. Many ISPs conduct their activities through WWW, which makes people have easier access to information on the web than in the libraries. Based on their technology and abundant information resources, many traditional libraries make use of Internet to develop themselves into digital libraries. However during their developing there arise many problems, which cause the phenomenon of "rich data, poor information" so that it is not easy for users to find much useable information immediately.Compared with the whole information space, information related to one's interest is very little. The fact that all the information itself stands in the same place for everyone may explain why one cannot get what he wants with ease. In recent studies one of the trends is that researchers attempt to personalize information service to find an efficient way for a particular user to search particular information. The key to doing this is to discover particular information each user needs to help him get what he wants easily.On this basis, the thesis takes into account the characteristics of Chinese information processing and makes use of multi-technology to come up with a personalized Chinese information service system for digital libraries to fulfill users' needs.The thesis first introduces the background of information service, the problems of existing system and some relative technology, and proposes solutions to thissystem. Then, considering the object's characteristics of this research, it puts forwardthe general modules of the system. The main body of this thesis is Chapter Three, in which we establish an executing model of personalized Chinese information service, and discuss the working processes of each part in detail. Li this part, the thesis first profiles semantic features of each document by employing Chinese information processing technology in order to change documents into the form which can be operated with the help of mathematical methods.Second, the thesis profiles each user's information needs by three ways: 1) accepting the information provided by the user himself; 2) watching the user's retrieval action; and 3) analyzing web server log. In this module, users are also classified into different categories according to their information needs. During the course of information recommending, the thesis combines content-based filtering with collaborative filtering to find particular information for each user, and it also presents different collaborative filtering strategies. In personalized information retrieval, the thesis takes advantage of the user's profile which describes particular user's information needs to filter useful information from the retrieval results for particular user and to enhance the accuracy of retrieval.Finally, the thesis makes some relevant experiments on the effects of information filtering in retrieval. Besides, it summarizes the thesis and provides suggestions for further research.