METHOD FOR STATISTIC VOICE LANGUAGE UNDERSTANDING USING GLOBAL TRIGGER FEATURE, AND MEDIUM RECORDING PROGRAM THEREOF

摘要

Statistical speech language understanding method using global triggering qualities according to the present invention is repeated global triggering according to quantitative criteria for comparing and evaluating the performance of previous speech language reader to which learning data is applied and speech language reader to which new global context information is added. Extracting a feature and adding the feature to a list of triggered features; Extracting global context information from the speech recognition sentence by using local context information and the trigger feature list in the speech recognition sentence; And synthesizing the local context information and the global context information in a speech language reader to output a semantic sentence.;According to the present invention, it is possible to automatically extract the triggering feature information to extract the global context information necessary to improve the performance of speech language understanding, thereby reducing the effort and cost required for parsing or writing rules, and easily applicable to various application areas. Provides scalability