Embedded grammar tags: advancing natural language interaction on the Web

Embedded grammar tags (EGTs) reflect Web page content by anticipating queries users might launch to retrieve that content. The grammars provide a unifying component for speech recognition engines, semantic Web page representation, and speech output generation. We propose a new framework that lets intelligent agents discover accurate, concise responses to natural language queries. This framework's backbone consists of embedded grammar tags that capture natural language queries. EGTs reflect information content in Web pages by anticipating the queries that users might launch to retrieve particular content.