[3O1-GS-13] AI application: Dialogue system and mind

[3O1-GS-13-04] Examination of a response generation method using attention mechanism considering speaker information

〇Masahiro Mizukami¹, Hiroaki Sugiyama¹, Hiromi Narimatsu¹, Tsunehiro Arimoto¹, Ryuichiro Higashinaka¹ (1. NTT Communication Science Laboratories)

Keywords:dialogue system

In a dialogue system, there is a demand to imitate a specific character to improve friendliness and amusingness.
Therefore, it is necessary to generate a response sentence that reflects the personality of the responding speaker, such as the way of speaking, personality, hobbies, and preferences.
In this study, we add an improvement to consider the speaker information in the attention mechanism, which is an effective mechanism in response generation, and examine a response generation model that reflects individuality.