Improving the closing sequences of interaction between human and robot through conversation analysis.

Area of Science:

• Human-Robot Interaction
• Computational Linguistics
• Social Robotics

Background:

• Current dialogue robots prioritize technical performance over user interaction quality.
• User experience and willingness for repeated interaction are crucial for daily-life robot adoption.
• Goffmanian insights and Conversation Analysis (CA) offer frameworks for understanding interactional dynamics.

Purpose of the Study:

• To develop a dialogue closing system for human-robot interaction.
• To enhance user perception of robots (likeability, politeness, satisfaction).
• To propose a recursive model for improving dialogue system design using CA.

Main Methods:

• Employing Conversation Analysis (CA) and Goffmanian insights.
• Developing and experimentally verifying a dialogue closing system.
• Qualitatively analyzing human-robot conversational data using CA.

Main Results:

• Robots with the dialogue closing system showed improved user perception.
• Users reported higher likeability, politeness, and dialogue satisfaction.
• A recursive model was proposed to enhance dialogue system design.

Conclusions:

• Interactional morality, specifically conversational closings, is key for human-like robot interaction.
• The recursive CA-based model enhances dialogue system quality by identifying necessary conversational properties.
• This approach facilitates the creation of robots that can coexist with humans.