Intelligent retrieval of dynamic networked information from mobile terminals using spoken natural language queries

This paper presents a working architecture for intelligent retrieval of dynamic networked information from mobile terminals using spoken natural language queries. A very efficient dynamic keyword extraction technique is used to extract automatically limited numbers of keywords to specify the huge amount of networked information, so as to reduce the very difficult problem of recognizing unconstrained spoken natural language queries into the tractable task of spotting limited number of keywords embedded in the voice queries. A vocabulary-flexible voice keyword spotting technique is used to handle the live, dynamic and ever-changing networked information. A client-server architecture is used to integrate the dynamic keyword extraction and voice keyword spotter modules under a mobile network environment. Very encouraging experimental results indicating the feasibility of the proposed approach have been obtained.