About 23 million documents match your query…

Abstract

A simple search of a large information space, such as the World Wide Web, often leaves the user to scan millions of hits. Relevance feedback is an information retrieval technique that can be used to make a search query more specific, so that its results are more manageable and useful. However, users tend not to take advantage of relevance feedback when systems provide it. I believe that this is because the process is badly represented at the search interface, and my thesis work to date has been aimed at facilitating relevance feedback by providing a visual representation of the user’s search context. An initial Java interface has been implemented, which will evolve in future to become a user-centred information workspace, supporting the search of a multimedia document collection.