To improve precision of interaction of an interactive system, which utilizes a speech to learn, by properly classifying and utilizing obtained knowledge for the interaction based upon characteristics of the knowledge.SOLUTION: A growth type interactive device is configured to perform speech interaction with a user and then to accumulate results of the speech interaction as knowledge, and comprises: a speech recognition part which performs speech recognition on an acquired user's uttered speech to create a text; an intention understanding part which analyzes utterance intention from the text created through the speech recognition of the speech recognition part by reference to intention understanding model data learnt from the intension understanding learning data; a response generation part which refers to a QA DB to generate a response text from the utterance intention analyzed by the intention understanding part; a knowledge extraction part which extracts knowledge from the text created through the speech recognition of the speech recognition part, the utterance intention, and the response text; and a knowledge classification part which classifies the knowledge extracted by the knowledge extraction part according to characteristics.SELECTED DRAWING: Figure 5
展开▼
