Conventional search engines such as Google or MSN provide access to Web information that can be acquired easily by crawling Web links. However, much valuable information is only accessible through source-specific search interfaces. Federated search provides access to this type of hidden Web contents by providing a single interface that connects to multiple source-specific search engines. My dissertation research addresses the three main research problems within federated search: resource representation, resource selection and results merging.
New algorithms have been proposed for estimating information source sizes, estimating distributions of relevant documents across information sources for resource selection, and merging document rankings returned by selected sources. Furthermore, a unified utility maximization framework is proposed to combine the range of solutions together to construct effective systems for different federated search applications. Empirical studies in a wide range of research environments and a real world prototype system under different operating conditions have demonstrated the effectiveness of the research. This new research, supported by a more theoretical foundation, better empirical results, and more realistic simulation of real world applications, substantially improves the state-of-the-art of federated search.