Empathetic Reflective Response Generation: Towards Conversation Models for Online Mental Health Support
Year: 2024
Type: conference-paper
Abstract: This paper presents an empathetic generative model that responds to online mental-health group dialogues with high-quality counseling-style reflection. Reflection is a crucial counseling technique that involves actively listening to the inter-locutor's thoughts and emotions, interpreting their meaning, and then articulating a concise, empathetic response to demonstrate understanding and promote deeper self-exploration. We discuss the importance of empathy in effective counseling approaches and study the impact of modeling multiple text-based empathetic factors on generating better reflective responses. Our model inte-grates various text-based empathetic components such as emotion and intent to enhance context and response representations, as well as three mechanisms for the communication of empathy in response. Empirical experiments demonstrate the significance of our approach through automated and human evaluations in terms of reflectiveness, empathy, coherence, and fluency.
Authors: Tootiya Giyahchi, Cornelia Pechmann, Ian Harris
Institution: University of California, Irvine
Language: English
Topic: Mental Health via Writing
Subfield: Social Psychology
Field: Psychology
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W4408146389