A retrieval method based on language model considering neighboring contents

The World Wide Web (WWW) has a massive number of Web pages, so that it is difficult for users to get useful information. In recent years, however, it is said that the probabilistic language model can help to improve retrieval accuracies of some kinds of search engines. The probabilistic language model has statistical background and can adapt previous text information retrieval model. However, we cannot directly adapt the probabilistic language model to Web search engines, because data on the Web has a hyperlink environment while the probabilistic language model can only text information on it. In this paper, we propose a novel approach for searching Web pages considering a Web page as well as its neighboring oneson the hyperlink environment.