With the rapid development of internet technology, more and more digital information is available for users. However, facing such large volumes of information makes users confused about how to find what they want. They get lost in this "information jungle." Some internet tools were developed to help users solve this problem, e.g. the Electronic Bulletin Board System, Newsgroups, and Gopher group information in a hierarchical form so that users can quickly find information that interests them.

more »

... W search engines such as Yahoo and Lycos provide a keyword-matching mechanism for users to search information that they are interested in. However, these tools cannot solve the problem completely. Although providing a way for users to look for existing information, they do not have the ability to deliver new information to users. Thus, users still have to look for information by these tools periodically so as not to miss something interesting. Selective dissemination of information (SDI) systems as described in Malone et al. (1987) were then developed to solve this problem. They filter information automatically and continuously deliver it to users [1]. In most of these systems, a user can describe his interests in a list of queries, called a user profile. And the system will then deliver information to him according to his profile. Typically, a query will contain some keywords that users are interested in, and the information that contains these keywords will be delivered to users. Although there is much nonEnglish information, most of the SDI systems support English only. Thus, multi-language filtering tools are needed. In this paper, we describe an SDI system that supports not only English but also Chinese. It follows the client-server model. Clients contact the server and express their interests. The server receives real-time news articles periodically and prepares relevant articles for clients. Clients can then browse information that has been prepared for them. In this paper we address the problem of the design and implementation of a real SDI system, especially user-interest modeling and Chinese filtering. The remainder of this paper is organized as follows. The next section briefly describes the 320 Abstract With the speedy growth of information quantity, people need a mechanism to discover automatically the information that interests them. Such a mechanism is called selective dissemination of information (SDI). Describes the design and implementation of an SDI system with the ability of delivering realtime, personalized news articles. In addition to delivering English news, it delivers Chinese articles also. Focuses on the problems that other researches seldom address. First, discusses how to store and delete news articles efficiently, then describes the user model to let users specify their interests. Finally, presents an efficient method to embed the ability to deliver Chinese as well as English news articles in the system.