Think Twice: A Human-like Two-stage Conversational Agent for Emotional Response Generation
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2023
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866914960877027328 |
|---|---|
| author | Qian, Yushan Wang, Bo Ma, Shangzhao Bin, Wu Zhang, Shuo Zhao, Dongming Huang, Kun Hou, Yuexian |
| author_facet | Qian, Yushan Wang, Bo Ma, Shangzhao Bin, Wu Zhang, Shuo Zhao, Dongming Huang, Kun Hou, Yuexian |
| contents | Towards human-like dialogue systems, current emotional dialogue approaches jointly model emotion and semantics with a unified neural network. This strategy tends to generate safe responses due to the mutual restriction between emotion and semantics, and requires rare emotion-annotated large-scale dialogue corpus. Inspired by the "think twice" behavior in human dialogue, we propose a two-stage conversational agent for the generation of emotional dialogue. Firstly, a dialogue model trained without the emotion-annotated dialogue corpus generates a prototype response that meets the contextual semantics. Secondly, the first-stage prototype is modified by a controllable emotion refiner with the empathy hypothesis. Experimental results on the DailyDialog and EmpatheticDialogues datasets demonstrate that the proposed conversational outperforms the comparison models in emotion generation and maintains the semantic performance in automatic and human evaluations. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2301_04907 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | Think Twice: A Human-like Two-stage Conversational Agent for Emotional Response Generation Qian, Yushan Wang, Bo Ma, Shangzhao Bin, Wu Zhang, Shuo Zhao, Dongming Huang, Kun Hou, Yuexian Computation and Language Human-Computer Interaction Towards human-like dialogue systems, current emotional dialogue approaches jointly model emotion and semantics with a unified neural network. This strategy tends to generate safe responses due to the mutual restriction between emotion and semantics, and requires rare emotion-annotated large-scale dialogue corpus. Inspired by the "think twice" behavior in human dialogue, we propose a two-stage conversational agent for the generation of emotional dialogue. Firstly, a dialogue model trained without the emotion-annotated dialogue corpus generates a prototype response that meets the contextual semantics. Secondly, the first-stage prototype is modified by a controllable emotion refiner with the empathy hypothesis. Experimental results on the DailyDialog and EmpatheticDialogues datasets demonstrate that the proposed conversational outperforms the comparison models in emotion generation and maintains the semantic performance in automatic and human evaluations. |
| title | Think Twice: A Human-like Two-stage Conversational Agent for Emotional Response Generation |
| topic | Computation and Language Human-Computer Interaction |
| url | https://arxiv.org/abs/2301.04907 |