A hybrid and regenerative model chat robot based on LSTM and attention model

Dongyang Gao; Junwu Zhu; Fudong Li; Yuequan Yang

doi:10.1117/12.2603769

28 October 2021 A hybrid and regenerative model chat robot based on LSTM and attention model

Dongyang Gao, Junwu Zhu, Fudong Li, Yuequan Yang

Proceedings Volume 11884, International Symposium on Artificial Intelligence and Robotics 2021; 118840H (2021) https://doi.org/10.1117/12.2603769
Event: International Symposium on Artificial Intelligence and Robotics 2021, 2021, Fukuoka, Japan

Abstract

Aiming at the situation that retrieval chat robot relies too much on predefined responses and the training requirements of generative chat robot are too high, a hybrid and regenerative model text chat robot based on LSTM and Attention-model is designed. Due to the retrieval model can only handle scenarios with predefined responses, and a generative model with strong learning ability will produce grammatical errors in certain scenarios. Therefore, firstly,doing text processing based on corpus, and then the retrieval model generates a candidate data set, and the candidate data set is trained by generating model to obtain the final model. The experimental comparison results show that the hybrid and regenerative model chat robot can effectively improve the model response quality compared to the single model chat robot, and accuracy improved by thirty percent.

Citation Download Citation

Dongyang Gao, Junwu Zhu, Fudong Li, and Yuequan Yang "A hybrid and regenerative model chat robot based on LSTM and attention model", Proc. SPIE 11884, International Symposium on Artificial Intelligence and Robotics 2021, 118840H (28 October 2021); https://doi.org/10.1117/12.2603769

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available