No abstract

SUMMARYWith the spreading use of the Web, there are increasing opportunities for information retrieval from the Internet. It is thus desirable to establish a method that can efficiently locate information required by a user from Web space, over which diverse information is scattered. This paper discusses a search scheme through links in Web space where the target search efficiency is maximized by utilizing the knowledge owned by the links. First, assuming that the probability that the target is present can be estimated from knowledge in Web pages, the estimation procedure is discussed. Then, the search process used to trace the links in Web space is modeled by a tree structure. Two problems are defined, namely, minimization of the retrieval cost and maximization of the number of retrieved pages. For each of these, an efficient search procedure is proposed. Lastly, the performance of the method is analyzed.