About

I am a senior scientist researcher at IAIC, A*STAR. Before IAIC was formed through the merger of I2R and IHPC on July 1, 2026, I worked at the Institute for Infocomm Research (I2R), A*STAR, where I contributed to MERaLiON-related research on multilingual, multimodal, and culturally grounded foundation models for Southeast Asia.

I received my Ph.D. from the NLP Research Group of Soochow University, China, under the supervision of Prof. Fang Kong and Prof. Guodong Zhou.

My research started from Discourse Analysis, Entailment Reasoning, and Summarization, and has recently expanded to multimodal LLMs, omni-perception models, perception-reasoning trade-offs, and long-video temporal reasoning. I am open to research collaborations on discourse-aware NLP, multimodal reasoning, culturally grounded AI, and temporal reasoning.

Email: Zhang_Longyin@a-star.edu.sg

Research Focus

Discourse-aware NLP

Discourse parsing, discourse-aware generation, summarization, and entailment reasoning beyond isolated sentences.

Top-down DRS Parsing Adversarial DRS Parser

Culturally-aware Multimodal LLMs

SEA-focused models and benchmarks that capture local visual, acoustic, linguistic, and cultural signals.

MERaLiON-Omni-MM-10B MERaLiON-Omni-GRPO-10B

Perception-Reasoning Trade-offs

Studying how System-2-style reasoning improves abstract tasks while causing temporal drift or visual over-interpretation.

arXiv:2602.23730

Cross-modal Temporal Reasoning

Event-aware long-video reasoning with physical-time grounding, audio-visual retrieval, and efficient temporal attention.

Projects

1. Singapore National Multimodal Large Language Model Programme: MERaLiON, 2025-. Research on multilingual and multimodal foundation models for Southeast Asia, focusing on visual recognition, multimodal reasoning, culturally grounded evaluation, and omni-perception capabilities. We are hiring!
2. Confidential applied AI research project with national-level partners, 2022-2025.
3. Neuro-Symbolic AI for Interpretable Natural Language Generation, A*STAR I2R, Singapore, 2022-2023. Research on interpretable generation with neural and symbolic reasoning components.

Demo

Here is a brief demo showcasing recent work on MERaLiON-Omni, including visual recognition, multilingual instruction following, and culturally-aware multimodal reasoning.

Research & Publications

Longyin Zhang, Shuo Sun, Yingxu He, Won Cheng Yi Lewis, Muhammad Huzaifah Bin Md Shahrin, Hardik Bhupendra Sailor, Jeremy Heng Meng Wong, and others. Unlocking Cognitive Capabilities and Analyzing the Perception-Logic Trade-off. arXiv 2026.

Chengwei Wei, Jihyuk Kim, Longyin Zhang, Shuo Chen, and Nancy F. Chen. InfoDensity: Rewarding Information-Dense Traces for Efficient Reasoning. arXiv 2026.

Yuanchen Shi, Longyin Zhang, Maodong Li, Yilin Zheng, Xin Wang, and Fang Kong. Simulating Human-Like Counseling: A Path-and Scenario-Guided Framework for Psychological Support Dialogue. In Proceedings of AAAI 2026.

Yuanchen Shi, Longyin Zhang, Guodong Zhou, and Fang Kong. Danger Depends on the Mind: A Theory-of-Mind Grounded Dataset and Model for Context-Dependent Dangerous Speech. In Proceedings of ACL 2026.

Longyin Zhang, Bowei Zou, and Aw Ai Ti. Enhancing Event-centric News Cluster Summarization via Data Sharpening and Localization Insights. In Proceedings of ACL 2025. A*STAR Research Highlight.

Maodong Li, Longyin Zhang, and Fang Kong. Multi-Hop Question Generation via Dual-Perspective Keyword Guidance. In Proceedings of ACL 2025.

Yuanchen Shi, Fang Kong, and Longyin Zhang. Impact of Stickers on Multimodal Sentiment and Intent in Social Media: A New Task, Dataset and Baseline. In Proceedings of ACMMM 2025.

Longyin Zhang, Bowei Zou, and Aw Ai Ti. Improving Multilingual Social Media Insights: Aspect-based Comment Analysis. arXiv 2025.

Longyin Zhang, Xin Tan, Fang Kong, and Guodong Zhou. Self-Trained and Self-Purified Data Augmentation for RST Discourse Parsing. In IEEE Transactions on Audio, Speech and Language Processing (TASLP).

Longyin Zhang. Research on Discourse Rhetorical Structure Analysis with Multilevel Knowledge Integration (融合多层次知识的篇章修辞结构解析研究). Outstanding Doctoral Dissertation of Jiangsu Province (China). 2019-2024

Longyin Zhang, Bowei Zou, Jacintha Wee Yun Yi, Ai Ti Aw. Comprehensive Abstractive Comment Summarization with Dynamic Clustering and Chain of Thought. In Proceedings of ACL 2024.

Longyin Zhang, Bowei Zou, Ai Ti Aw. Empowering Tree-structured Entailment Reasoning: Rhetorical Perception and LLM-driven Interpretability. In Proceedings of LREC-COLING 2024.

Longyin Zhang, Xin Tan, Peifeng Li, Fang Kong, and Guodong Zhou. Top-Down Text-Level Discourse Rhetorical Structure Parsing with Bidirectional Representation Learning. In Journal of Computer Science and Technology (JCST). 2023.

Yuqing Xing, Longyin Zhang, Fang Kong, Guodong Zhou. Discourse Parsing Enhanced by Discourse Dependence Perception. In Proceedings of AACL 2022.

Xin Tan, Longyin Zhang, Fang Kong, and Guodong Zhou. Towards Discourse-aware Document-level Neural Machine Translation. IJCAI 2022, pages 4383-4389.

Longyin Zhang, Fang Kong, and Guodong Zhou. Adversarial Learning for Discourse Rhetorical Structure Parsing. In Proceedings of ACL 2021, pages 3946-3957.

Longyin Zhang, Yuqing Xing, Fang Kong, Peifeng Li, and Guodong Zhou. A Top-Down Neural Architecture towards Text-Level Parsing of Discourse Rhetorical Structure. In Proceedings of ACL 2020, pages 6386-6395.

Longyin Zhang, Xin Tan, Fang Kong, and Guodong Zhou. EDTC: A Corpus for Discourse-Level Topic Chain Parsing. In Proceedings of EMNLP2021 findings, pages 1304-1312.

Xin Tan, Longyin Zhang, and Guodong Zhou. Coupling Context Modeling with Zero Pronoun Recovering for Document-Level Natural Language Generation. EMNLP 2021, pages 2530-2540.

Longyin Zhang, Fang Kong, and Guodong Zhou. Syntax-Guided Sequence to Sequence Modeling for Discourse Segmentation. In Proceedings of NLPCC 2020, pages 95-107.

Longyin Zhang, Xin Tan, Fang Kong, and Guodong Zhou. A Recursive Information Flow Gated Model for RST-style Text-level Discourse Parsing. In Proceedings of NLPCC 2019, pages 231-241. 2019.

Longyin Zhang, Cheng Sun, Xin Tan, and Fang Kong. RST Discourse Parsing With Tree-Structured Neural Networks. In Proceedings of CWMT, Communications in Computer and Information Science, vol 954, pages 15-26. 2019.

Jinfeng Wang, Longyin Zhang, and Fang Kong. Multi-Level Cohesion Information Modeling for Better Written and Dialogue Discourse Parsing. NLPCC 2021, pages 40-52.

戴倩雯, 张龙印, 孔芳. 融合依存关系和篇章修辞关系的事件时序关系识别. 模式识别与人工智能, 32(12):1100-1106. 2019.

Xin Tan, Longyin Zhang, Deyi Xiong, and Guodong Zhou. Hierarchical Modeling of Global Context for Document-Level Neural Machine Translation. EMNLP 2019, pages 1576-1585.

谭新, 邝少辉, 张龙印, 熊德意. 融入汉字笔画序列的神经机器翻译. 厦门大学学报（自然科学版）, 58(2):164-169. 2019.