SocialSim: Towards Socialized Simulation of Emotional Support Conversation

2025-04-11

AAAI Conference on Artificial Intelligence

AI Summary

Existing emotion-support conversation (ESC) simulation methods overlook social dynamics, resulting in limited authenticity and scalability. To address ESC data scarcity, this paper proposes the first synthetic framework explicitly modeling dual-role social behaviors: (1) on the seeker side, it introduces persona-bank-driven social self-disclosure; (2) on the supporter side, it integrates socially grounded chain-of-thought reasoning to enhance response logicality; and (3) it employs LLM-based controllable generation coupled with multi-stage quality filtering to construct SSConv—a high-fidelity synthetic corpus. Experiments demonstrate that models trained on SSConv achieve state-of-the-art performance on ESC tasks, outperforming crowdsourced baselines in both automated and human evaluations. This work is the first to systematically incorporate social self-disclosure and social awareness into ESC simulation, overcoming the fundamental limitation of conventional language models—namely, their neglect of the interactive, socially embedded nature of support conversations.

Technology Category

• Cognitive Modeling & Cognitive Systems: Social Cognition And Interaction
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Application Domains: Humanities & Computational Social Science

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Search and Retrieval-Augmented AI: Assisted, interactive, and conversational search

Abstract

Emotional support conversation (ESC) helps reduce people's psychological stress and provide emotional value through interactive dialogues. Due to the high cost of crowdsourcing a large ESC corpus, recent attempts use large language models for dialogue augmentation. However, existing approaches largely overlook the social dynamics inherent in ESC, leading to less effective simulations. In this paper, we introduce SocialSim, a novel framework that simulates ESC by integrating key aspects of social interactions: social disclosure and social awareness. On the seeker side, we facilitate social disclosure by constructing a comprehensive persona bank that captures diverse and authentic help-seeking scenarios. On the supporter side, we enhance social awareness by eliciting cognitive reasoning to generate logical and supportive responses. Building upon SocialSim, we construct SSConv, a large-scale synthetic ESC corpus of which quality can even surpass crowdsourced ESC data. We further train a chatbot on SSConv and demonstrate its state-of-the-art performance in both automatic and human evaluations. We believe SocialSim offers a scalable way to synthesize ESC, making emotional care more accessible and practical.

Problem

• Simulating emotional support with social dynamics
• Reducing cost of creating ESC corpus
• Enhancing quality of synthetic ESC data

Innovation

• Integrates social disclosure and awareness
• Constructs comprehensive persona bank
• Enhances cognitive reasoning for responses