TY - GEN
T1 - MateBot
T2 - 15th International Conference on Green, Pervasive, and Cloud Computing, GPC 2020
AU - Wang, Ziqi
AU - Guo, Bin
AU - Wang, Hao
AU - Cui, Helei
AU - He, Yang
AU - Yu, Zhiwen
N1 - Publisher Copyright:
© 2020, Springer Nature Switzerland AG.
PY - 2020
Y1 - 2020
N2 - The virtual bot is one of the hot topics in artificial intelligence, where most of the current studies focus on chatbots. Nevertheless, the context-sensitive virtual bot, especially with rich human-like interactions (e.g., appearance change, context-aware narration/recommendation) regarding the ambient changes (e.g., location, focused scene) through various built-in sensors, would have broader application. Towards this direction, we propose MateBot, a human-like, context-sensitive virtual bot, which supports harmonious human-computer interaction on smartphones. The design of MateBot consists of three parts. First, a context sensing network is used to recognize the input background information and face information, and modify the appearance of the virtual bot through the conversion of the encoding network. Second, a human-like bot appearance generation network can generate a virtual bot image with a human-like appearance through the GAN network and modify the appearance of the virtual bot with context-sensitive information. Third, a personalized conversation network is devised to communicate with human users. Furthermore, we apply MateBot to the intelligent travel scenario to justify its practicality, and the experiment results show that the bot can better increase the user’s sense of substitution and improve the communication efficiency between human users and virtual bots.
AB - The virtual bot is one of the hot topics in artificial intelligence, where most of the current studies focus on chatbots. Nevertheless, the context-sensitive virtual bot, especially with rich human-like interactions (e.g., appearance change, context-aware narration/recommendation) regarding the ambient changes (e.g., location, focused scene) through various built-in sensors, would have broader application. Towards this direction, we propose MateBot, a human-like, context-sensitive virtual bot, which supports harmonious human-computer interaction on smartphones. The design of MateBot consists of three parts. First, a context sensing network is used to recognize the input background information and face information, and modify the appearance of the virtual bot through the conversion of the encoding network. Second, a human-like bot appearance generation network can generate a virtual bot image with a human-like appearance through the GAN network and modify the appearance of the virtual bot with context-sensitive information. Third, a personalized conversation network is devised to communicate with human users. Furthermore, we apply MateBot to the intelligent travel scenario to justify its practicality, and the experiment results show that the bot can better increase the user’s sense of substitution and improve the communication efficiency between human users and virtual bots.
KW - Context-sensitive
KW - MateBot
KW - Virtual bot
UR - http://www.scopus.com/inward/record.url?scp=85097848333&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-64243-3_21
DO - 10.1007/978-3-030-64243-3_21
M3 - 会议稿件
AN - SCOPUS:85097848333
SN - 9783030642426
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 273
EP - 287
BT - Green, Pervasive, and Cloud Computing - 15th International Conference, GPC 2020, Proceedings
A2 - Yu, Zhiwen
A2 - Becker, Christian
A2 - Xing, Guoliang
PB - Springer Science and Business Media Deutschland GmbH
Y2 - 13 November 2020 through 15 November 2020
ER -