USING FEEDBACK IN CONTENT- IMAGE METASEARCH

he proliferation of online documents and search engines in the pastfew years has motivated the development of metasearch engines,which act as gateways that link users automatically and transpar-ently to multiple and maybe competing search engines. Currentmetasearch engines use a variety of techniques (see the sidebar, “RelatedWork”), though all of them work with the fundamental searching unit oftext. Automated visual information retrieval (VIR) systems work withimage features—such as colors, textures, and shapes—in combination withtext and other related information to query the increasing store of imagedata available on the Web.