Xiaoying Song

square.png

Hi, I’m Xiaoying, a Ph.D. Candidate in Data Science at the University of North Texas, advised by Dr. Lingzi Hong. Prior to joining UNT, I received my M.S. degree in Information Science from Central China Normal University in 2023. My work has appeared at NLP and information science venues including EMNLP, NAACL, COLING, and COLM, as well as in journals such as Journal of Information Science, Journal of Informetrics, and Machine Learning.

I work on Large Language Models and Natural Language Processing, centered on two questions: how agents collaborate with humans and with each other, and how they personalize to the user in front of them. I design multi-agent pipelines with humans in the loop and adapt model behavior to each user, from reading level to safety, using RAG, RL, PEFT/LoRA, and quantization-aware training. I target high-stakes applications where trust decides whether a system ships: content moderation, misinformation response, and crisis communication. My work spans three areas:

Human–Agent Collaboration

Multi-turn conversation optimization and multi-agent coordination for complex, evolving tasks.

Multi-agent RAG against health misinformation COLM 2025 Forecasting hater reactions to counterspeech NAACL 2025

Personalized Alignment

User-centered dialogue systems and safety alignment that adapts to who the user is.

Literacy-controlled counterspeech with RAG-RL EMNLP 2025 On-demand safety alignment relaxation Preprint 2026

Computational Social Science

Language technologies for healthier online discourse: countering misinformation and supporting crisis response.

Consistent crisis response via dynamic fusion EMNLP 2025 Human likeness of AI-generated counterspeech COLING 2025 Outcome-constrained LLMs for countering hate speech EMNLP 2024

News

May 06, 2026 Our team won second place in the CLEF CheckThat! 2026 competition. 🥈
Apr 23, 2026 I delivered a tutorial for the 2026 ASIS&T IDEA program on AI-assisted communication. 🧑‍🏫
Apr 13, 2026 I defended my Ph.D. qualifying exam and became a Ph.D. candidate. 🎓
Nov 10, 2025 “Speaking at the Right Level” got Honorable Mention in ASIS&T SIG-SM Student Competition. 🏅
Aug 20, 2025 Two papers were accepted to EMNLP 2025. 🎉

Selected Publications

  1. Preprint
    PALETTE.png
    Palette: A Modular, Controllable, and Efficient Framework for On-demand Authorized Safety Alignment Relaxation in LLMs
    Qitao Tan, Xiaoying Song, Arman Akbari, Arash Akbari, Yanzhi Wang, Xiaoming Zhai, Lingzi Hong, Zhen Xiang, Jin Lu, and Geng Yuan
    arXiv preprint arXiv:2605.24154, 2026 PDF
    Proposes Palette, a framework that selectively relaxes LLM refusal behavior in authorized professional domains while preserving safety elsewhere, combining refusal-direction search, lightweight adaptation, and parameter merging for multi-domain composition without retraining.
    safety alignment refusal behavior model merging controllability
  2. EMNLP
    Speak.png
    Speaking at the Right Level: Literacy-Controlled Counterspeech Generation with RAG-RL
    Xiaoying Song, Anirban Saha Anik, Dibakar Barua, Pengcheng Luo, Junhua Ding, and Lingzi Hong
    In Conference on Empirical Methods in Natural Language Processing, 2025 PDF
    Proposes Controlled-Literacy, a RAG plus reinforcement learning framework that generates counterspeech to health misinformation tailored to the audience’s health literacy level, producing more accessible and preferred counterspeech than baselines.
    readability alignment health misinformation RAG reinforcement learning
  3. EMNLP
    Crisis.png
    A Dynamic Fusion Model for Consistent Crisis Response
    Xiaoying Song, Anirban Saha Anik, Eduardo Blanco, Vanessa Frias-Martinez, and Lingzi Hong
    In Conference on Empirical Methods in Natural Language Processing, 2025 PDF
    Introduces a metric for style consistency in automated crisis responses and a two-stage fusion approach that scores candidate responses and integrates them at the instance level, improving both response quality and stylistic uniformity across datasets.
    crisis communication style consistency text generation model fusion
  4. COLM
    multi_agent.png
    Multi-Agent Retrieval-Augmented Framework for Evidence-Based Counterspeech Against Health Misinformation
    Xiaoying Song*, Anirban Saha Anik*, Elliott Wang, Bryan Wang, Bengisu Yarimbas, and Lingzi Hong
    In Conference on Language Modeling, 2025 PDF
    Presents a multi-agent retrieval-augmented framework that combines static and dynamic evidence to generate counterspeech against health misinformation, outperforming baselines in politeness, relevance, informativeness, and factual accuracy.
    dialogue system health misinformation multi-agent systems RAG
  5. NAACL
    hatespeech.png
    Echoes of Discord: Forecasting Hater Reactions to Counterspeech
    Xiaoying Song, Sharon Lisseth Perez, Xinchen Yu, Eduardo Blanco, and Lingzi Hong
    In Nations of the Americas Chapter of the Association for Computational Linguistics, 2025 PDF
    Studies whether counterspeech leads haters to reenter conversations and whether reentry stays hateful, introducing ReEco, a dataset of triple-turn Reddit conversations, and linking counterspeech language to hater reactions.
    response effectiveness hate speech user behavior prediction