摘要 |
A device and a method for providing a universal conversation service by using an information state and an example-based inference model according to a category of dialogue are provided to increase quality of the dialogue by inferring the dialogue and making a system response to all kinds of user questions intelligently based on information obtained from entire dialogue, which is included in a purpose and general category. A semantic interpreter(22) analyzes a user intention by interpreting the meaning of an input query in a classified category. A response inference engine(24) infers a present system intention for the user intention classified into each category. A system state controller(23) generates present system state information from the previous user/system intention and the accumulated system state information. A response template generator(28) generates a system response template for a target category from the preset user/system intention. A system response generator() generates the system response from the system response template for the target and general category.
|