09:00–09:15OpeningOpening RemarksSophia AnaniadouThe University of Manchester
Morning
Daniel Preoțiuc-Pietro is the engineering manager of the AI Search Services Group at Bloomberg, New York City. His group develops the search models, tools and infrastructure that power both low-latency search and conversational use cases with access to hundreds of millions of unstructured documents across the Bloomberg Terminal.
His research encompasses topics in NLP, LLMs and Computational Social Science, where he published over 50 top-tier publications. Primary current research areas include introducing new task setups, applications and NLP resources inspired by industry use cases. Personal research interests focus on analyzing the pragmatic, social and temporal dimensions of text, particularly from social media, with cross-disciplinary applications in psychology, law, political science and journalism. This research has been featured extensively in popular press including The Washington Post, BBC and New Scientist.
He co-founded and co-organizes the Natural Legal Language Processing (NLLP) workshop series collocated with *ACL conferences and serves as Industry Track Chair for EACL 2027. Prior to joining Bloomberg, he was a postdoctoral research fellow at the University of Pennsylvania and obtained a PhD in the NLP Group at the University of Sheffield, UK.
09:45–10:30Oral Session IFinancial Language AgentsSix papers, five minutes each, then joint Q&A
- How LLM Financial Agents Respond to Stated Affect Depends on Their Assigned RoleDa Chen Lian, Yu-Lin Chang, Yen-Hsi Huang, Shu-Kai Hsieh
- When Helpful Agents Cause Financial Harm: Evaluating LLM Agent Safety in Executable Economic Worlds RemoteYuzhu Ling, Jiale Han, Benyou Wang
- DealTrace: Localizing and Intervening on Failures in Agentic Financial ReasoningAlina Hyk
- MortarBench: Evaluating Mortgage Loan Origination AgentsMatthew Toles, Yunan Lu, Manav Munjal, Bojun Liu, Yuanhao Deng, Stephanie Selig, Derek Rindner, Cheng Li, Zhou Yu
- Theatron: A Benchmark for Evaluating Persona Persistence in Long-Horizon LLM Decision-MakingYuyan Wang, Yuechen Jiang, Zhiwei Liu, Jimin Huang, Sophia Ananiadou
- AlphaMemo: Structured Search-Process Memory for Self-Evolving Alpha Mining AgentsHang Yu, Zifan Zheng, Jeff Z. Pan, Tongliang Liu, Zhiyong Wang, Fengxiang He
10:30Coffee break
Yohei Seki is a Professor at the Institute of Library, Information and Media Science, University of Tsukuba, Japan, and was selected as one of the University of Tsukuba's Best Faculty Members for 2025. He received his Ph.D. from the Graduate University for Advanced Studies (SOKENDAI) in 2005. His research interests include natural language processing, information access, and citizen sentiment analysis, with a recent focus on the analysis of corporate ESG disclosures. He co-organised the NTCIR Multilingual Opinion Analysis Task from 2006 to 2010, and served as a Japanese-side organiser of the ML-ESG shared tasks and SemEval-2025 Task 6 (PromiseEval). His work was a runner-up for the Best Paper Award at ICADL 2022. He also co-edited the international smart city standards ISO/IEC 30146 (2019) and ISO/IEC 30145-3 (2020).
Prayag Tiwari is an Associate Professor at Halmstad University. He received his Ph.D. from the University of Padova, Italy. He has previously worked as a Postdoctoral Researcher at Aalto University, Finland, and as a Marie Curie Researcher at the University of Padova, Italy. Recipient of the 2022 Best Paper Award for a publication in Neural Networks and recognized for multiple Highly Cited Papers. Author of numerous publications in top-tier journals and conferences, including Nature Communications, NeurIPS, ICML, CVPR, ICLR, AAAI, ACL, EMNLP, TMLR, IJCV, PAMI, ACM MM, ECCV, WWW, CIKM. Editor for journals such as Transactions on Machine Learning Research, Neural Networks, Quantum Machine Intelligence and IEEE Transactions on Fuzzy Systems, and Area Chair in NeurIPS 2026/27, AAAI 2026/27 and ICLR 2026/27.
Bhaskarjit Sarmah is an AI research leader specializing in Agentic AI and financial services. He currently leads AI Research for Financial Services at Domyn, focusing on Knowledge Graphs, RAG, agentic AI, LLM evaluation, and responsible AI. Previously, he led AI research initiatives at BlackRock focused on applying Generative AI and machine learning to investment and risk management. He is also an educator, researcher, and frequent speaker on AI and financial technology.
12:30–14:00Lunch & PostersLunch and Poster SessionHall Hall accepted papers, including the orals · see the papers
Afternoon
Jacob Chanyeol Choi, Ph.D. is Co-founder and CEO of LinqAlpha, which builds multi-agent AI for institutional investment research. He holds a Ph.D. in Electrical Engineering and Computer Science from MIT, where his work on neuromorphic computing and AI hardware appeared in Nature, Nature Materials and Nature Electronics and has been cited over 5,800 times. He co-organized the first LLMs and Generative AI for Finance workshop at ICAIF, and his team publishes on financial NLP evaluation, including FinDER and FinAgentBench. LinqAlpha is building open benchmarks for investment research agents with Google DeepMind's Kaggle Benchmarks team, and its retrieval model ranked first on the HuggingFace MTEB leaderboard in 2024.
Qianqian Xie is a Professor at the School of Artificial Intelligence, Wuhan University. She was an Associate Research Scientist at Yale University. She is the Secretary of SIG-FinTech. Her research focuses on Financial NLP and BioNLP, with over 60 papers published in conferences and journals, such as NeurIPS, ACL, KDD, ICLR, Nature Communications. She has served as (senior) area chair for NeurIPS, ICLR, ACL, KDD, EMNLP, COLING, LREC, and ARR, and as General Co-Chair for COLING 2025 FinNLP workshop. She received the NIH K99/R00 Pathway to Independence Award, and CCL 2024 Outstanding English Paper Award. She is a co-organizer of several financial shared tasks such as FinNLPAgentScen@IJCAI-2024.
15:00–15:30Shared TaskFinEval: Overview and AwardsOverview and subtask awardswith short presentations from the winning teams
15:30Coffee break
16:00–16:45Oral Session IIFinancial Text Understanding and ReasoningSix papers, five minutes each, then joint Q&A
- FinLCR: Scaling Financial Long-Context Reasoning via Curriculum Reinforcement LearningWenjie Qiu, Yong Xie, Karan Aggarwal, Aitzaz Ahmad, Stephen Lau
- Characterizing Ambiguity in Financial Document Question AnsweringKelsey Rook, Clare Arrington, Tomek Strzalkowski, Razvan Bunescu
- Diagnosing Fine-Grained Inconsistency Classification in Financial Disclosure TextAman Kumar, Lasitha Vidyaratne, Dipanjan Dipak Ghosh, Arnab Chakrabarti, Ahmed K. Farahat
- What Ensemble Brier Scores Hide: Evidence Routing at a Fixed Article-Exposure Budget in LLM Prediction-Market Forecasting RemoteAvinash Goutham Aluguvelly
- GRPO for Financial Advice Generation: Outperforming Commercial LLMs under CATE EvaluationOfir Ben Shoham, Shrutendra Harsola, Vignesh Subrahmaniam, Shravan Mohan, Yakov Gazman, Oded Vainas
- ShiJianBench: From Dialogue to Decision for Long-Horizon Evaluation of Investment AdvisorsJie Gong, Maowei Jiang, Zhiwei Liu, Yang Qiao, Wenxi Wu, Mengxi Xiao, Enze Zhang, Ziyan Kuang, Yankai Chen, Caishuang Huang, Meng Zhou, Xiku Du, Xue Liu, Guojun Xiong, Min Peng, Qianqian Xie, Sophia Ananiadou
16:45–17:15PanelWhy FinNLP — and Who Is It For?Moderated by Chung-Chi Chenwith the invited speakers
Why does financial NLP merit being a field in its own right, what problems is it uniquely placed to address, and who ultimately benefits? Beyond the usual research and industry framing, the panel turns to the broader public value of the field: financial transparency, reducing information asymmetry, supporting regulators and journalists, protecting consumers, and making complex financial information accessible. And the harder question: if much of FinNLP helps financial institutions operate or invest more effectively, what should the wider contribution of the field be?
17:15–17:30ClosingDomyn Best Paper Award and Closing RemarksYongjae Lee
The Domyn Best Paper Award is presented to three papers, selected from the oral papers by an independent review panel. FinEval subtask awards are presented during the shared task session.
Evening
19:00–21:00ReceptionFinNLP NightSponsored by Bloombergvenue to be announced
