INTELLIGENT RETRIEVAL OF DYNAMIC NETWORKED INFORMATION FROM MOBILE TERMINALS USING SPOKEN NATURAL-LANGUAGE QUERIES

This paper presents a working architecture for intelligent retrieval o f dynamic networked information from mobile terminals using spoken nat ural language queries. A very efficient dynamic keyword extraction tec hnique 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 la nguage queries into the tractable task of spotting limited number of k eywords embedded in the voice queries. A vocabulary-flexible voice key word spotting technique is used to handle the live, dynamic and ever-c hanging networked information. A client-server architecture is used to integrate the dynamic keyword extraction and voice keyword spotter mo dules under a mobile network environment. Very encouraging experimenta l results indicating the feasibility of the proposed approach have bee n obtained.