CubanSea: Cluster-Based Visualization of Search Results

In recent years, there has been a move toward supporting the human element of Web search beyond a simple query box and a ranked list of search results. In this paper, we present a prototype implementation of an approach to Web search based on fuzzy clustering and visualization. CubanSea presents the searcher with topics that are automatically discovered within the search space. A visual encoding of the cluster membership supports the searcher in understanding the relationship between the search results and the fuzzy clusters. Clusters can be selected for evaluation, and filtering operations allow the searcher to further manipulate and explore the search results set. The system is designed to support searchers when their information needs are ill-defined, vague, or ambiguous.