Publications

@inproceedings{fang_social_2025,

title = {Social Simulation for Everyday Self-Care: Design Insights from Leveraging VR, AR, and LLMs for Practicing Stress Relief},

author = {A. Fang and H. Chhabria and A. Maram and H. Zhu},

url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-105005770377&doi=10.1145%2F3706598.3713115&partnerID=40&md5=828b06008a1409e9dc32425e568f4f33},

doi = {10.1145/3706598.3713115},

isbn = {9798400713958 (ISBN); 9798400713941 (ISBN)},

year  = {2025},

date = {2025-01-01},

booktitle = {Conf Hum Fact Comput Syst Proc},

publisher = {Association for Computing Machinery},

abstract = {Stress is an inevitable part of day-to-day life yet many find themselves unable to manage it themselves, particularly when professional or peer support are not always readily available. As self-care becomes increasingly vital for mental well-being, this paper explores the potential of social simulation as a safe, virtual environment for practicing in-the-moment stress relief for everyday social situations. Leveraging the immersive capabilities of VR, AR, and LLMs to create realistic interactions and environments, we developed eight interactive prototypes for various social stress related scenarios (e.g. public speaking, interpersonal conflict) across design dimensions of modality, interactivity, and mental health guidance in order to conduct prototype-driven semi-structured interviews with 19 participants. Our qualitative findings reveal that people currently lack effective means to support themselves through everyday stress and perceive social simulation - even at low immersion and interaction levels - to fill a gap for practical, controlled training of mental health practices. We outline key design needs for developing social simulation for self-care needs, and identify important considerations including risks of trauma from hyper-realism, distrust of LLM-recommended timing for mental health recommendations, and the value of accessibility for self-care interventions. © 2025 Elsevier B.V., All rights reserved.},

keywords = {design, Design insights, Language Model, Large language model, large language models, Mental health, Peer support, Professional supports, Self-care, Social simulations, Speed dating, Virtual environments, Virtual Reality, Well being},

pubstate = {published},

tppubtype = {inproceedings}

}

@inproceedings{zhang_self-emotion_2024,

title = {Self-Emotion Blended Dialogue Generation in Social Simulation Agents},

author = {Q. Zhang and J. Naradowsky and Y. Miyao},

editor = {T. Kawahara and V. Demberg and S. Ultes and K. Inoue and S. Mehri and D. Howcroft and K. Komatani},

url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-105017744334&doi=10.18653%2Fv1%2F2024.sigdial-1.21&partnerID=40&md5=f185cfb5554eabfa85e6e956dfe6848e},

doi = {10.18653/v1/2024.sigdial-1.21},

isbn = {9798891761612 (ISBN)},

year  = {2024},

date = {2024-01-01},

pages = {228–247},

publisher = {Association for Computational Linguistics (ACL)},

abstract = {When engaging in conversations, dialogue agents in a virtual simulation environment may exhibit their own emotional states that are unrelated to the immediate conversational context, a phenomenon known as self-emotion. This study explores how such self-emotion affects the agents' behaviors in dialogue strategies and decision-making within a large language model (LLM)-driven simulation framework. In a dialogue strategy prediction experiment, we analyze the dialogue strategy choices employed by agents both with and without self-emotion, comparing them to those of humans. The results show that incorporating self-emotion helps agents exhibit more human-like dialogue strategies. In an independent experiment comparing the performance of models fine-tuned on GPT-4 generated dialogue datasets, we demonstrate that self-emotion can lead to better overall naturalness and humanness. Finally, in a virtual simulation environment where agents have discussions on multiple topics, we show that self-emotion of agents can significantly influence the decision-making process of the agents, leading to approximately a 50% change in decisions. © 2025 Elsevier B.V., All rights reserved.},

keywords = {Agent behavior, Agents, Computational Linguistics, Decision making, Decisions makings, Dialogue generations, Dialogue strategy, Emotional state, Language Model, Model-driven, Natural language processing systems, Simulation framework, Social psychology, Social simulations, Speech processing, Virtual Reality, Virtual simulation environments},

pubstate = {published},

tppubtype = {inproceedings}

}