Classified displays of web search results.

Most text retrieval systems return a ranked list of results in response to a user's search request. Such lists can be long and overwhelming. Furthermore, results on different topics or different aspects of the same topic are intermixed in the list requiring users to sift through a long undifferentiated list to find items of interest. We have been exploring the use of automatic text classification techniques combined with novel interface ideas to allow users to quickly focus in on results of interest. Our approach combines the advantages of human knowledge in an initial classification stage with the broad coverage available with text retrieval systems. In a series of user studies we developed and evaluated several interfaces for structuring search results in order to better understand the cognitive processes that lead to effective analysis of search results. There are two key aspects to our work that we describe in more detail: 1) automatic text classification algorithms for quickly and accurately tagging ew content, and 2) novel interface to support structured search.