대화분석을 통한 지능형 대화시스템의 비교분석

초록

This paper examines interactional data drawn from human-computer interface that was developed by computer engineers. A total of 1,860 cases of conversational exchanges were collected from 62 participants who interacted with service bots in three domains (Schedule, Contact and Weather). The research followed the principle of conversation analysis (CA) by tracing line-by-line progression of interactional exchanges. The analysis produced 19 different categories of codes through a qualitative research from which three most common problems emerged. The finding offers a useful insight into some notable challenges in developing dialog systems that are to interact with human participants. Since the findings are drawn from methodological approaches in social sciences, they provide somewhat different solutions to engineering problems in human-computer interface. The manuscript also includes introductory descriptions of the four papers that are published in this special volumes on human-computer interaction.

키워드