We describe an approach for constructing search spaces that consist of highly relevant web pages using similarities between the contents of linked web pages to represent their link distance. Because this approach was generally time consuming, we designed an approach for our cooperative web architecture that would shorten the search time. We applied our approach to an actual web link space and evaluated it by comparing it with an approach using fixed link distances. We focused on whether such a filtering approach was effective in reducing the search space of neighboring web searches . KEYWORDS Web information retrieval, search space, similarity, cooperative Web architecture, link graph.