IEEE Robotics and Automation Letters · 2021 · 19 citations · 15 references
EngineeringComputer SciencePsychologyHuman-robot InteractionRecommender SystemsRoboticsConversational RecommendationEmbodied InteractionHuman-robot CommunicationHuman-robot Interaction EvaluationConsumer Food PreferencesRetail RoboticsRobot-mediated CommunicationSocially Interactive Robots
Recent developments in conversational technologies have attracted researchers to study their applications in recommending items through conversations. It is considered that physical robots, rather than virtual ones, are effective in situations in which robots talk about items near participants. However, in real situations, robots may be required to recommend items that are present but invisible in the scene of the communication. In this study, we conducted a field experiment in a shopping mall to investigate the effects of robot embodiment on recommendation tasks. The robots recommended a dish after talking to participants about their food preferences. We developed a conversational recommendation system and implemented it using physical and virtual robots. The field experiment was conducted in a shopping mall; the visitors were encouraged to participate. The experiment lasted a total of 99 hours (9 hours per day for 11 days) inside and in front of a food court. Although no significant difference in the behavioral aspect was confirmed, the results obtained from 272 conversations suggested that having physical bodies enhanced the satisfaction and agreement with the robots’ recommendations.
15
Embodiment in Socially Interactive Robots
Eric Deng, Bilge Mutlu, Maja J. Matarić · Foundations and Trends in Robotics · 2019 · 217 citations · Full text
Neural personalized response generation as domain adaptation
Weinan Zhang, Qingfu Zhu, Yifa Wang et al. · World Wide Web · 2018 · 92 citations