AI 스피커에 적용을 위한 대화 함축의 해석 범주와 과정 분석 연구 - 초·중등학생의 대화 함축의 해석과 오류 해소 양상을 중심으로 -

초록

The purpose of this study is to conduct a survey among elementary and middle school students, to identify the interactive agents realized by AI speakers, and reveal an understanding of conversational implications and how to build data by checking the variables and processes of implicit interpretation. The survey helped identify the variables and processes for understanding the categories and implications of dialogue, and examined the factors involved in resolving communication problems arising due to the misinterpretation of such implications. We constructed ten realistic conversational implications and asked the respondents to answer three multiple-choice questions regarding each implication. The results showed the interpretation of conversational implications to be sufficiently limited, such that specific responses were provided for seven of the ten dialogue situations. Among the factors considered in the interpretation of conversational implications, “the situation in which the conversation took place” was the most frequent, followed by “understanding the meaning of words.” When a misinterpretation of the conversational implications led to a communication disorder, the most common method used by the respondents to resolve it was “asking.” Significant differences were observed between the two genders and different grade levels. This study is significant because it shows that if data are constructed in a way that provides clues about the “conversation situation” in the user’s utterance, it is possible to create an interactive agent that understands the implied information, even at a rudimentary level.

키워드