Improving Multi-turn Emotional Support Dialogue Generation with Lookahead Strategy Planning

Abstract

Providing Emotional Support (ES) to soothe people in emotional distress is an essential capability in social interactions. Most existing research on building ES conversation systems only considers single-turn interactions with users, which is over-simplified. In comparison, multi-turn ES conversation systems can provide ES more effectively, but face several new technical challenges, including: i) how to conduct support strategy planning that could lead to the best supporting effects; ii) how to dynamically model the user's state. In this paper, we propose a novel system named MultiESC to address these issues. For strategy planning, drawing inspiration from the A* search algorithm, we propose lookahead heuristics to estimate the future user feedback after using particular strategies, which helps to select strategies that can lead to the best long-term effects. For user state modeling, MultiESC focuses on capturing users' subtle emotional expressions and understanding their emotion causes. Extensive experiments show that MultiESC significantly outperforms competitive baselines in both strategy planning and dialogue generation. Our codes are available at https: //github.com/lwgkzl/MultiESC.

AdMISC: Advanced Multi-Task Learning and Feature-Fusion for Emotional Support Conversation

“…On the other hand, Peng et al [11] incorporated seeker emotional feedback information for dialogue strategy selection. Xu et al [12] employed a prior knowledge method in predicting dialogue strategy labels, and Cheng et al [13] considered forward-looking heuristic strategy planning and selection.…”

SoulChat: Improving LLMs’ Empathy, Listening, and Comfort Abilities through Fine-tuning with Multi-turn Empathy Conversations

“…ChatGLM (Zeng et al, 2023), SparkDesk 3 , as presented in Appendix F. It may be due to the lack of large-scale multi-turn empathy conversation datasets for fine-tuning stage, especially in the field of Chinese mental health or emotional support. EMPATHETICDIALOGUES (Rashkin et al, 2019) and ESConv (Liu et al, 2021) are two English empathy conversation datasets that is used for developing emotional support conversation (ESC) systems, e.g MISC (Tu et al, 2022), GLHG (Peng et al, 2022), MultiESC (Cheng et al, 2022), FADO (Peng et al, 2023) and etc. On the one hand, these models may rely on annotated empathy strategies and emotions of users during the training or inference stage, which means that building large-scale similar datasets for fine-tuning LLMs is difficult.…”