RESPONSE EVALUATION OF SPOKEN DIALOGUE PROCESSING METHOD USING INDUCTIVE LEARNING BASED AMOUNT OF INFORMATION

摘要

To be effective, a spoken dialogue system must be able to process and accurately respond to conversation as well as adapting to background noise, interjections and unnecessary vocabulary. To do so the system needs to obtain information from the dialogue automatically. Therefore we propose a learning method using AoI(Amount of Information). We define the rule of spoken dialogue based on AoI. In this paper, we describe the experiment and results which show a 44% improvement over the Inductive Learning method.