Liyan Xu

About Me

I am currently a Senior Research Scientist at WeChat AI, Tencent, focusing on Natural Language Processing (NLP) and Large Language Models (LLM). I graduated with PhD of Computer Science from Emory University, advised by Dr. Jinho Choi, and co-advised by Dr. Fei Liu in 2023.

Research Interests

My current research interests focus around LLM Memory and Agentic Context Comprehension.

Looking forward to research collaboration and discussions.

News

• [06/2023]: started Research Scientist @ WeChat AI.
• [05/2023]: completed my PhD journey @ Emory!
• [05/2022]: started Applied Scientist Intern @ Amazon Product Graph.
• [05/2021]: started Applied Scientist Intern @ Amazon Alexa ASR.
• [10/2020]: started research collaboration @ NEC Labs America.
• [08/2020]: winning team of the 2020 Amazon Alexa Prize Chatbot Competition.

Selected Publications

(*: Equal Contribution; †: Project Lead)

Language Modeling

• Rank Is the Bottleneck: Measuring and Training Pluggable Memory for LLMs.
  Liyan Xu, Jiangnan Li, Zhiqiang Yuan, Zexi Jia, Jinchao Zhang, Jie Zhou
  arXiv Soon.

• Think Less, Locally: Local Preference Learning for Reasoning Compression.
  Muyao Yuan, Liyan Xu^*†, Weizhan Zhang, Muyan Jiao, Jiangnan Li, Jinchao Zhang, Jie Zhou
  arXiv Soon.

• How Far Ahead Do LLMs Plan? Uncovering the Latent Horizon in Chain-of-Thought Reasoning
  Liyan Xu, Mo Yu, Fandong Meng, Jie Zhou
  [ICML’26] The 43rd International Conference on Machine Learning.

• ReFreeKV: Towards Threshold-Free KV Cache Compression
  Xuanfan Ni^*, Liyan Xu^*†, Chenyang Lyu, Longyue Wang, Mo Yu, Lemao Liu, Fandong Meng, Jie Zhou, Piji Li^†
  [ACL’26 Findings] The 64th Annual Meeting of the Association for Computational Linguistics.

• RescoreBERT: Discriminative Speech Recognition Rescoring with BERT
  Liyan Xu, Yile Gu, Jari Kolehmainen, Haidar Khan, Ankur Gandhe, Ariya Rastrow, Andreas Stolcke, Ivan Bulyko
  [ICASSP’22] 2022 IEEE International Conference on Acoustics, Speech and Signal Processing.

• Boosting Cross-Lingual Transfer via Self-Learning with Uncertainty Estimation
  Liyan Xu, Xuchao Zhang, Xujiang Zhao, Haifeng Chen, Feng Chen, Jinho Choi
  [EMNLP’21] The 2021 Conference on Empirical Methods in Natural Language Processing.

Context Comprehension

• PRELUDE: A Benchmark Designed to Require Global Comprehension and Reasoning over Long Contexts
  Mo Yu^*, Tsz Ting Chung^*, Chulun Zhou^*, Tong Li^*, Rui Lu^*, Jiangnan Li^*, Liyan Xu^*, Haoshu Lu, Ning Zhang, Jing Li, Jie Zhou
  arXiv preprint 2508.09848

• SitEmb-v1.5: Improved Context-Aware Dense Retrieval for Semantic Association and Long Story Comprehension
  Junjie Wu, Jiangnan Li, Yuqing Li, Lemao Liu, Liyan Xu, Jiwei Li, Dit-Yan Yeung, Jie Zhou, Mo Yu
  [ACL’26] The 64th Annual Meeting of the Association for Computational Linguistics.

• ComoRAG: A Cognitive-Inspired Memory-Organized RAG for Stateful Long Narrative Reasoning
  Juyuan Wang^*, Rongchen Zhao^*, Wei Wei, Yufeng Wang, Mo Yu, Jie Zhou, Jin Xu, Liyan Xu^†
  [AAAI’26] The 40th AAAI Conference on Artificial Intelligence.

• Fine-Grained Modeling of Narrative Context: A Coherence Perspective via Retrospective Questions
  Liyan Xu, Jiangnan Li, Mo Yu, Jie Zhou
  [ACL’24] The 62nd Annual Meeting of the Association for Computational Linguistics.

• Zero-Shot Cross-Lingual Machine Reading Comprehension via Inter-Sentence Dependency Graph
  Liyan Xu, Xuchao Zhang, Bo Zong, Yanchi Liu, Wei Cheng, Jingchao Ni, Haifeng Chen, Liang Zhao, Jinho Choi
  [AAAI’22] The 36th AAAI Conference on Artificial Intelligence.

Representation Learning

• Dense Retrievers Can Fail on Simple Queries: Revealing The Granularity Dilemma of Embeddings
  Liyan Xu, Zhenlin Su, Mo Yu, Jiangnan Li, Fandong Meng, Jie Zhou
  [EMNLP’25 Findings] The 2025 Conference on Empirical Methods in Natural Language Processing.

• Towards Open-World Product Attribute Mining: A Lightly-Supervised Approach
  Liyan Xu, Chenwei Zhang, Xian Li, Jingbo Shang, Jinho Choi
  [ACL’23] The 61st Annual Meeting of the Association for Computational Linguistics.

Information Extraction & Discourse

• Identifying Factual Inconsistencies in Summaries: Grounding Model Inference via Task Taxonomy
  Liyan Xu^*, Zhenlin Su^*, Mo Yu, Jin Xu, Jinho Choi, Jie Zhou, Fei Liu
  [EMNLP’24 Findings] The 2024 Conference on Empirical Methods in Natural Language Processing.

• SIG: Speaker Identification in Literature via Prompt-Based Generation
  Zhenlin Su, Liyan Xu^†, Jin Xu^†, Jiangnan Li, Mingdu Huangfu
  [AAAI’24] The 38th AAAI Conference on Artificial Intelligence.

• Modeling Task Interactions in Document-Level Joint Entity and Relation Extraction
  Liyan Xu, Jinho Choi
  [NAACL’22] The 2022 Conference of the North American Chapter of the Association for Computational Linguistics.

• Online Coreference Resolution for Dialogue Processing: Improving Mention-Linking on Real-Time Conversations
  Liyan Xu, Jinho Choi
  [*SEM’22] The 11th Joint Conference on Lexical and Computational Semantics.

• Adapted End-to-End Coreference Resolution System for Anaphoric Identities in Dialogues
  Liyan Xu, Jinho Choi
  The CODI-CRAC 2021 Shared Task on Anaphora, Bridging, and Discourse Deixis in Dialogue.

• Revealing the Myth of Higher-Order Inference in Coreference Resolution
  Liyan Xu, Jinho Choi
  [EMNLP’20] The 2020 Conference on Empirical Methods in Natural Language Processing.

Professional Services

• Program Committee Member: ARR Rolling (ACL, EMNLP, NAACL), AAAI, NeurIPS, ICML, ICLR

Other Activities

I enjoy wandering in nature with my camera in hand (my gallery).