As we build towards developing interactive systems that can recognize human emotional states and respond to individual needs more intuitively and empathetically in more personalized and context-aware computing time. This is especially important regarding mental health support, with a rising need for immediate, non-intrusive help tailored to each individual. Individual mental health and the complex nature of human emotions call for novel approaches beyond conventional proactive and reactive-based chatbot approaches. In this position paper, we will explore how to create Chatbots that can sense, interpret, and intervene in emotional signals by combining real-time facial expression analysis, physiological signal interpretation, and language models. This is achieved by incorporating facial affect detection into existing practical and ubiquitous passive sensing contexts, thus empowering them with the capabilities to the ubiquity of sensing behavioral primitives to recognize, interpret, and respond to human emotions. In parallel, the system employs cognitive-behavioral therapy tools such as cognitive reframing and mood journals, leveraging the therapeutic intervention potential of Chatbots in mental health contexts. Finally, we propose a project to build a system that enhances the emotional understanding of Chatbots to engage users in chat-based intervention, thereby helping manage their mood.

翻译：随着我们致力于开发能够识别人类情感状态、更直观和共情地响应个体需求的交互系统，计算正朝着更加个性化和情境感知的方向发展。这在心理健康支持领域尤为重要，因为对即时、非侵入性且针对个体的帮助需求日益增长。个体的心理健康和人类情感的复杂性要求超越传统主动式和反应式聊天机器人方法的新颖途径。在本立场文件中，我们将探讨如何通过结合实时面部表情分析、生理信号解读和语言模型，创建能够感知、解释并干预情感信号的聊天机器人。这是通过将面部情感检测融入现有实用且普遍存在的被动感知情境中实现的，从而赋予它们利用无处不在的行为基元感知能力来识别、解释和响应人类情感。同时，该系统采用认知行为疗法工具，如认知重构和情绪日记，利用聊天机器人在心理健康情境中的治疗干预潜力。最后，我们提出一个项目，旨在构建一个增强聊天机器人情感理解能力的系统，以吸引用户参与基于聊天的干预，从而帮助他们管理情绪。