研究与发表

@article{wen2026metactrl,
  title = {{MetaCtrl}: Your Large Language Models Can Reason Better and More Concisely with a Metacognitive Controller},
  author = {Wen, Zhibin and Han, Tao and Bai, Lei and Li, Can and Xu, Yang},
  journal = {arXiv preprint arXiv:2609.37304},
  year = {2026},
  eprint = {2609.37304},
  archivePrefix = {arXiv},
  primaryClass = {cs.AI},
  url = {https://arxiv.org/abs/2609.37304}
}

@article{an-xu-2026-fiscore,
  title = "Teaching {LLMs} to Abstain via Fine-Grained Semantic Confidence Reward ({FiSCoRe})",
  author = "An, Hao  and Xu, Yang",
  year = "2026",
  journal = "arXiv preprint arXiv:2510.24020",
  url = "https://arxiv.org/abs/2510.24020",
}

@inproceedings{li-xu-2026-uidpo,
  title = {Uniform Information Density-Based Preference Optimization},
  author = {Li, Yongyuan and Xu, Yang},
  booktitle = {Conference on Language Modeling},
  year = {2026},
  url = {https://openreview.net/forum?id=7PEYAPbjtN},
  note = {Accepted for presentation at COLM 2026}
}

@inproceedings{ou-etal-2026-identifying,
    title = "Identifying the Periodicity of Information in Natural Language",
    author = "OU, Yulin  and
      Wang, Yu  and
      Xu, Yang  and
      Buschmeier, Hendrik",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Proceedings of the 64th Annual Meeting of the {A}ssociation for {C}omputational {L}inguistics (Volume 1: Long Papers)",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2026.acl-long.52/",
    doi = "10.18653/v1/2026.acl-long.52",
    pages = "1161--1175",
    ISBN = "979-8-89176-390-6",
    abstract = "Recent theoretical advancement of information density in natural language has brought the following question on desk: To what degree does natural language exhibit periodicity pattern in its encoded information? We address this question by introducing a new method called AutoPeriod of Surprisal (APS). APS adopts a canonical periodicity detection algorithm and is able to identify any significant periods that exist in the surprisal sequence of a single document. By applying the algorithm to a set of corpora, we have obtained the following interesting results: Firstly, a considerable proportion of human language demonstrates a strong pattern of periodicity in information; Secondly, new periods that are outside the distributions of typical structural units in text (e.g., sentence boundaries, elementary discourse units, etc.) are found and further confirmed via harmonic regression modeling. We conclude that the periodicity of information in language is a joint outcome from both structured factors and other driving factors that take effect at longer distances. The advantages of our periodicity detection method and its potentials in LLM-generation detection are further discussed."
}

@inproceedings{liu-etal-2026-think,
    title = "Think in Sentences: Explicit Sentence Boundaries Enhance Language Model{'}s Capabilities",
    author = "Liu, Zhichen  and
      Li, Yongyuan  and
      Xu, Yang",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Proceedings of the 64th Annual Meeting of the {A}ssociation for {C}omputational {L}inguistics (Volume 1: Long Papers)",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2026.acl-long.104/",
    doi = "10.18653/v1/2026.acl-long.104",
    pages = "2275--2288",
    ISBN = "979-8-89176-390-6",
    abstract = "Researchers have explored ways to improve large language models (LLMs)' capabilities via dummy token insertion in contexts. However, existing works focus solely on the dummy tokens themselves, but failed to leverage the inherent sentence-level structure of natural language. This is a critical oversight, as LLMs acquire linguistic capabilities through exposure to human-generated texts, which are inherently structured at the sentence level. Motivated by the gap, we proposed a method that inserts delimiters at sentence boundaries. Our method not only integrates dummy tokens into contexts, but also enables LLMs with sentence-by-sentence processing behavior during reasoning. Two approaches are proposed: (1). In-context learning and (2). Supervised fine-tuning are experimented from 7B LLMs to 600B Deepseek-V3. Experimental results demonstrate consistent improvements in various tasks, with notable gains of up to 7.7{\%} on GSM8k and 12.5{\%} on DROP. Furthermore, LLMs fine-tuned via our strategy further incorporate sentence awareness into their inner representations. Our work establishes a simple yet effective technique for enhancing LLM{'}s capabilities, offering promising directions for cognitive-inspired LLM enhancement paradigm."
}

@inproceedings{wang-etal-2026-investigating,
    title = "Investigating the Representation of Backchannels and Fillers in Fine-tuned Language Models",
    author = "Wang, Yu  and
      Lao, Leyi  and
      Huang, Langchu  and
      Skantze, Gabriel  and
      Xu, Yang  and
      Buschmeier, Hendrik",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Proceedings of the 64th Annual Meeting of the {A}ssociation for {C}omputational {L}inguistics (Volume 1: Long Papers)",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2026.acl-long.241/",
    doi = "10.18653/v1/2026.acl-long.241",
    pages = "5319--5348",
    ISBN = "979-8-89176-390-6",
    abstract = "Backchannels and fillers are important linguistic expressions in dialogue, but often treated as ``noise'' to be bypassed in modern transformer-based language models. Our work studies the representation of them in language models using three fine-tuning strategies. The models are trained on three dialogue corpora in English and Japanese, where backchannels and fillers are preserved and annotated, to investigate how fine-tuning can help LMs learn their representations. We first apply clustering analysis to the learnt representation of backchannels and fillers, and have found increased silhouette scores in representations from fine-tuned models, which suggests that fine-tuning enables LMs to distinguish the nuanced semantic variation in different backchannel and filler use. We also use natural language generation (NLG) metrics and qualitative analysis to confirm that the utterances generated by fine-tuned language models resemble human-produced utterances more closely. Our findings suggest the potentials of transforming general LMs into conversational LMs that are more capable of producing human-like languages adequately."
}

@inproceedings{liu-etal-2026-wavedetect,
    title = "{W}ave{D}etect: Robust Framework for Machine-Generated Text Detection via Wavelet Transform",
    author = "Liu, Zhichen  and
      Qin, Kaitong  and
      He, Linhan  and
      Xu, Yang",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {ACL} 2026",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2026.findings-acl.424/",
    doi = "10.18653/v1/2026.findings-acl.424",
    pages = "8712--8727",
    ISBN = "979-8-89176-395-1",
    abstract = "As Large Language Models asymptotically approach human-level fluency in natural language generation, solely relying on surface-level semantic artifacts for detecting LLM-generated texts has become increasingly precarious. Existing detectors often falter when facing three critical challenges: adversarial perturbations, cross-domain shifts, and the rapid temporal evolution of the foundation model. To address these issues, we propose , a novel framework that reformulates text detection as a signal processing task within the time-frequency domain. Unlike previous methods that analyze static token probability distributions, models the generated output as a probability signal, upon which a differentiable Continuous Wavelet Transform is applied to convert them into learnable spectral representations. This process reveals the intrinsic ``spectral fingerprints'' in machine-generated texts{--}patterns that remain invisible in time domain. Comprehensive evaluations on three well-curated datasets (RAID, EvoBench, and Domain-Shift) show that our method achieves a new state-of-the-art. It not only achieves superior accuracy but also exhibits remarkable robustness against sophisticated attacks, generalization across out-of-distribution topics and unseen evolving LLMs. Our results validate the efficacy of spectral analysis as a promising paradigm for LLM-generated texts detection."
}

@inproceedings{an-etal-2026-purging,
    title = "Purging the Gray Zone: Latent-Geometric Denoising for Precise Knowledge Boundary Awareness",
    author = "An, Hao  and
      Lou, Yibin  and
      Guo, Jiayi  and
      Xu, Yang",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {ACL} 2026",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2026.findings-acl.122/",
    doi = "10.18653/v1/2026.findings-acl.122",
    pages = "2562--2576",
    ISBN = "979-8-89176-395-1",
    abstract = "Large language models (LLMs) often exhibit hallucinations due to their inability to accurately perceive their own knowledge boundaries. Existing abstention fine-tuning methods typically partition datasets directly based on response accuracy, causing models to suffer from severe label noise near the decision boundaries and consequently exhibit high rates of abstentions or hallucinations. This paper adopts a latent space representation perspective, revealing a ``gray zone'' near the decision hyperplane where internal belief ambiguity constitutes the core performance bottleneck. Based on this insight, we propose the \textbf{GeoDe} (\textbf{Geo}metric \textbf{De}noising) framework for abstention fine-tuning. This method constructs a truth hyperplane using linear probes and performs ``geometric denoising'' by employing geometric distance as a confidence signal for abstention decisions. This approach filters out ambiguous boundary samples while retaining high-fidelity signals for fine-tuning. Experiments across multiple models (Llama3, Qwen3) and benchmark datasets (TriviaQA, NQ, SciQ, SimpleQA) demonstrate that GeoDe significantly enhances model truthfulness and demonstrates strong generalization in out-of-distribution (OOD) scenarios. Code is available at \url{https://github.com/Notbesidemoon/GeoDe}."
}

@inproceedings{zheng2026memory,
  title = {Evaluating Memory Capability in Continuous Lifelog Scenario},
  author = {Zheng, Jianjie and Liu, Zhichen and Shen, Zhanyu and Qu, Jingxiang and Chen, Guanhua and Wang, Yile and \textbf{Xu}, \textbf{Yang} and Liu, Yang and Cheng, Sijie},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  url = {https://arxiv.org/abs/2604.11182},
  note = {(CCF-A)}
}

@inproceedings{he-etal-2026-eye,
    title = "Eye Movement Features Can Predict Human Preferences on Machine-Generated Texts",
    author = "He, Xiaoshan  and
      Liu, Xiaoqun  and
      He, Haodong  and
      Wang, Yu  and
      Xu, Yang",
    editor = "T.Y.S.S., Santosh  and
      Rodriguez, Juan Diego  and
      de Gibert, Ona",
    booktitle = "Proceedings of the 64th Annual Meeting of the {A}ssociation for {C}omputational {L}inguistics (Volume 4: Student Research Workshop)",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2026.acl-srw.42/",
    doi = "10.18653/v1/2026.acl-srw.42",
    pages = "470--486",
    ISBN = "979-8-89176-393-7",
    abstract = "Eye movement offers valuable insights into human visual attention during assessment of machine-generated texts, yet existing research and resources in this area are limited. To bridge this gap, we introduce Gaze Responses for Evaluating AI Texts (GREAT), a comprehensive dataset capturing human eye-movement features during screen reading of passages generated by large language models (LLMs). The dataset includes raw eye-movement recordings, reading-time measurements, and post-reading evaluations for LLM-generated passage pairs, alongside rigorous validation metrics. The collected eye-movement features demonstrate strong explanatory power in predicting text quality. When integrated with negative log-likelihood (NLL), a commonly used metric for evaluating text quality, it substantially enhances model performance across all standard statistical criteria. These findings demonstrate that eye-movement can act as an effective source of information that complements probabilistic metrics, for the task of automatic text quality assessment. The full dataset and some processing code are publicly available at \url{https://github.com/qwurd231/GREAT}."
}

@inproceedings{liu2025face2,
  title = {Evaluating Text Generation Quality Using Spectral Distances of Surprisal},
  author = {Liu, Zhichen and Li, Yongyuan and Xu, Yang and Wang, Yu and Yuan, Yingfang and Yang, Zuhao},
  booktitle = {Findings of the Association for Computational Linguistics: EMNLP 2025},
  year = {2025},
}

@inproceedings{nguyen-xu-2025-reasoning,
  title = "Reasoning for Translation: Comparative Analysis of Chain-of-Thought and Tree-of-Thought Prompting for {LLM} Translation",
  author = "Nguyen, Lam  and Xu, Yang",
  booktitle = "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 4: Student Research Workshop)",
  month = jul,
  year = "2025",
  address = "Vienna, Austria",
  publisher = "Association for Computational Linguistics",
  url = "https://aclanthology.org/2025.acl-srw.17/",
  pages = "259--275",
}

@inproceedings{10.1145/3757376.3771409,
  series = {SA Technical Communications '25},
  location = {},
  numpages = {5},
  articleno = {30},
  booktitle = {Proceedings of the SIGGRAPH Asia 2025 Technical Communications},
  doi = {10.1145/3757376.3771409},
  url = {https://doi.org/10.1145/3757376.3771409},
  address = {New York, NY, USA},
  publisher = {Association for Computing Machinery},
  isbn = {9798400721366},
  year = {2025},
  title = {ReChar: Revitalising Characters with Structure Preserved and User-Specified Aesthetic Enhancements},
  author = {Yang, Zhongyu and Song, Junhao and Luo, Zhang and Yang, Zuhao and Xu, Yang and Lan, Jingfen and Zhang, Yonghan and Pang, Wei and Song, Siyang and Yuan, Yingfang},
}

@inproceedings{xu-etal-2024-detecting,
  title = {Detecting Subtle Differences between Human and Model Languages Using Spectrum of Relative Likelihood},
  author = {Xu, Yang  and Wang, Yu  and An, Hao  and Liu, Zhichen  and Li, Yongyuan},
  booktitle = {Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing},
  year = {2024},
  address = {Miami, Florida, USA},
  publisher = {Association for Computational Linguistics},
  url = {https://aclanthology.org/2024.emnlp-main.564},
  doi = {10.18653/v1/2024.emnlp-main.564},
  pages = {10108--10121},
}

@inproceedings{wang2024much,
  title = {How Much Does Nonverbal Communication Conform to Entropy Rate Constancy?: A Case Study on Listener Gaze in Interaction},
  author = {Wang, Yu and Xu, Yang and Skantze, Gabriel and Buschmeier, Hendrik},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2024},
  year = {2024},
}

@article{yang2024face,
  title = {Face: Evaluating natural language generation with fourier analysis of cross-entropy},
  author = {Yang, Zuhao and Yuan, Yingfang and Xu, Yang and Zhan, Shuo and Bai, Huajun and Chen, Kefan},
  journal = {Advances in Neural Information Processing Systems},
  volume = {36},
  year = {2024},
}

@inproceedings{xu-cheng-2023-spontaneous,
  title = {Spontaneous gestures encoded by hand positions improve language models: An Information-Theoretic motivated study},
  author = {Xu, Yang  and
  Cheng, Yang},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2023},
  month = {jul},
  year = {2023},
  address = {Toronto, Canada},
  publisher = {Association for Computational Linguistics},
  url = {https://aclanthology.org/2023.findings-acl.600},
  pages = {9409--9424},
}

@inproceedings{xu2022gestures,
  title={Gestures Are Used Rationally: Information Theoretic Evidence from Neural Sequential Models},
  author={Xu, Yang and Cheng, Yang and Bhatia, Riya},
  booktitle={Proceedings of the 29th International Conference on Computational Linguistics},
  pages={134--140},
  year={2022}
}

@article{GAYNES2022104386,
  title = {Hierarchical data visualization of experimental erythrocyte aggregation employing cross correlation and optical flow applications},
  journal = {Microvascular Research},
  pages = {104386},
  year = {2022},
  issn = {0026-2862},
  doi = {https://doi.org/10.1016/j.mvr.2022.104386},
  author = {Bruce I. Gaynes and Mark B. Shapiro and Abel Saju Augustine and Yang Xu and Yang Lin and Parisa Mirbod and Robert Dieter and Yang Cheng and Mengren Wu and Harish Venkataraman and Yuan Gao and Plamen Petrov and Jie Xu},
}

@article{xu2021global,
  title={Global divergence and local convergence of utterance semantic representations in dialogue},
  author={Xu, Yang},
  journal={Proceedings of the Society for Computation in Linguistics},
  volume={4},
  number={1},
  pages={116--124},
  year={2021}
}

@incollection{xu2021book,
  title = {Historical changes in semantic weights of sub-word units},
  author = {Xu, Yang and Zhang, Zheng-sheng},
  booktitle = {Computational Approaches to Semantic Change},
  editor = {Jatowt, Adam and Tahmasebi, Nina and Xu, Yang and Borin, Lars},
  year = {2021},
  pages = {169--187},
  publisher = {Language Science Press}
}

@article{kelly2020sentence,
  title={Which Sentence Embeddings and Which Layers Encode Syntactic Structure?},
  author={Kelly, M Alex and Xu, Yang and Calvillo, Jes{\'u}s and Reitter, David},
  year={2020},
  publisher={PsyArXiv}
}

@article{xu2019linguistic,
  title={Linguistic alignment is affected more by lexical surprisal rather than social power},
  author={Xu, Yang and Cole, Jeremy and Reitter, David},
  journal={Proceedings of the Society for Computation in Linguistics},
  volume={2},
  number={1},
  pages={349--352},
  year={2019}
}

@article{xu2019cogsciabs,
  title={},
  author={Xu, Yang and Zhang, Jiasheng and Reitter, David},
  journal={Proceedings of The 41st Annual Meeting of the Cognitive Science Society},
  volume={},
  number={},
  pages={},
  address={Montreal, Canada},
  year={2019}
}

@article{xu2019lchange,
  title={Treat the Word As a Whole or Look Inside? Subword Embeddings Model Language Change and Typology},
  author={Xu, Yang and Zhang, Jiasheng and Reitter, David},
  journal={Proceedings of the 1st International Workshop on Computational Approaches to Historical Language Change},
  volume={},
  number={},
  pages={},
  address={Florence, Italy},
  year={2019}
}

@article{xu_reitter_2018_cognition,
  title = {Information density converges in dialogue: Towards an information-theoretic model},
  journal = {Cognition},
  volume = {170},
  pages = {147--163},
  year = {2018},
  doi = {https://doi.org/10.1016/j.cognition.2017.09.018},
  url = {http://www.sciencedirect.com/science/article/pii/S0010027717302615},
  author = {Xu, Yang and Reitter, David}
}

@inproceedings{xu2018power,
  title={Not that much power: Linguistic alignment is influenced more by low-level linguistic features rather than social power},
  author={Xu, Yang and Cole, Jeremy and Reitter, David},
  booktitle={Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  volume={1},
  pages={601--610},
  year={2018}
}

@inproceedings{xu_reitter_2017_spectral,
  title={Spectral Analysis of Information Density in Dialogue Predicts Collaborative Task Performance},
  author={Xu, Yang and Reitter, David},
  booktitle={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  pages={623--633},
  year={2017},
  address={Vancouver, Canada}
  publisher = {Association for Computational Linguistics}
}

@InProceedings{xu_reitter2016complexity,
  author    = {Xu, Yang  and  Reitter, David},
  title     = {Convergence of Syntactic Complexity in Conversation},
  booktitle = {Proceedings of the 54th {Annual} {Meeting} of the {Association} for {Computational} {Linguistics} ({Volume} 2: {Short} {Papers})},
  year      = {2016},
  address   = {Berlin, Germany},
  publisher = {Association for Computational Linguistics},
  pages     = {443--448},
  url       = {http://anthology.aclweb.org/P16-2072}
}

@InProceedings{xu_reitter2016entropy,
  author    = {Xu, Yang and Reitter, David},
  title     = {Entropy converges between dialogue participants: Explanations from an information-theoretic perspective},
  booktitle = {Proceedings of the 54th {Annual} {Meeting} of the {Association} for {Computational} {Linguistics} ({Volume} 1: {Long} {Papers})},
  year      = {2016},
  address   = {Berlin, Germany},
  publisher = {Association for Computational Linguistics},
  pages     = {537--546},
  url       = {http://www.aclweb.org/anthology/P16-1051}
}

@InProceedings{xu_reitter2015evaluation,
  title={An evaluation and comparison of linguistic alignment measures},
  author={Xu, Yang and Reitter, David},
  journal={Proceedings of Cognitive Modeling and Computational Linguistics},
  pages={58--67},
  year={2015},
  publisher = {Association for Computational Linguistics},
  address = {Denver, CO}
}

@incollection{ororbia_ii_error-correction_2015,
  series = {Lecture {Notes} in {Computer} {Science}},
  title = {Error-correction and aggregation in crowd-sourcing of geopolitical incident information},
  volume = {9021},
  url = {http://www.david-reitter.com/pub/ororbia2015crowd.pdf},
  booktitle = {Social {Computing}, {Behavioral} {Modeling} and {Prediction}},
  publisher = {Springer},
  author = {Ororbia II, Alexander G. and Xu, Yang and D'Orazio, Vito and Reitter, David},
  editor = {Agarwal, Nitin and Xu, Kevin and Osgood, Nathaniel},
  year = {2015},
  note = {10.1007/978-3-319-16268-3 47},
  pages = {381--387}
}