Jie Zhang
Author directoryOther people with similar names: Jie Zhang, Jie Zhang, Jie Zhang, Jie Zhang, Jie Zhang, Jie Zhang, Jie Zhang, Jie Zhang
Unverified author pages with similar names: Jie Zhang
2026
ReimburseBench: Benchmarking LMMs as Full-Workflow Auditors for Financial Internal Control
Libin Zhan | Wuyou Mao | Xiang Li | Enci Zhang | Qi Yang | Shujian Zhu | Yingtian Hu | Wentao Ye | Haobo Wang | Ziman Li | Jie Zhang | Wang Dong | Jun Chen
Findings of the Association for Computational Linguistics: EMNLP 2026
Libin Zhan | Wuyou Mao | Xiang Li | Enci Zhang | Qi Yang | Shujian Zhu | Yingtian Hu | Wentao Ye | Haobo Wang | Ziman Li | Jie Zhang | Wang Dong | Jun Chen
Findings of the Association for Computational Linguistics: EMNLP 2026
Expense reimbursement internal control is a high-stakes corporate finance task that combines multimodal, visually rich evidence (e.g., receipts and approval forms), firm-specific policy text, and multi-step compliance reasoning. No existing benchmark evaluates large multimodal models (LMMs) in this domain. We introduce ReimburseBench, the first benchmark targeting full-workflow multimodal audit reasoning in corporate internal control. It comprises 151 expert-constructed reimbursement workflows, each packaging an application, receipts and supporting documents, and a firm-specific policy, evaluated by a process-level rubric grounded in twelve auditing logic points. Benchmarking nine frontier and locally deployable systems, we find the strongest attains only 72.5%, and no model exceeds 47.7% on hard workflows. Beyond this aggregate score, risk-aware, process-level evaluation surfaces three structural findings: (i) a persistent gap between binary admissibility judgment and precise quantitative computation; (ii) a systematic lenient bias in monetary predictions that poses deployment risk; and (iii) a performance pattern consistent with compositional-generalization difficulties under realistic workflow load. We provide ReimburseBench as a testbed for advancing LMMs toward trustworthy financial systems. Data and code: https://github.com/ikerzhan/ReimburseBench.
Rubric-Guided Process Reward for Stepwise Model Routing
Shenghao Ye | Yu Guo | Zhengheng Li | Qiang Huang | Shuangwu Chen | Wang Luo | Rui Qian | Jie Zhang | Tao Zhang | Dong Jin | Xiaobin Tan | Quan Zheng | Jianyang
Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
Shenghao Ye | Yu Guo | Zhengheng Li | Qiang Huang | Shuangwu Chen | Wang Luo | Rui Qian | Jie Zhang | Tao Zhang | Dong Jin | Xiaobin Tan | Quan Zheng | Jianyang
Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
Stepwise model routing improves the efficiency of Large Reasoning Models (LRMs) by assigning each reasoning step to a suitable model. Recent methods formulate routing as a sequential decision process and train the router with reinforcement learning. However, although they model routing as a process, they still supervise the router with outcome rewards. Such rewards only reflect final answer correctness and fail to evaluate intermediate routing decisions, which can weaken performance and generalization. To address this gap, we propose RoRo, a rubric-guided process reward framework for stepwise model routing. RoRo first collects diverse routing trajectories and constructs preference pairs based on outcome, cost, and process quality. It then trains a Rubricor to generate a query-specific evaluation rubric and a Judge to score routing trajectories under this rubric through alternating optimization. The resulting process rewards are combined with outcome rewards to optimize the routing policy via GRPO. Experiments on five reasoning benchmarks under both same-family and cross-family settings show that RoRo consistently outperforms strong baselines and achieves better accuracy and cost trade-offs.
From Past To Path: Masked History Learning for Next-Item Prediction in Generative Recommendation
Kaiwen Wei | Kejun he | Xiaomian Kang | Jie Zhang | Ymyang | Li Jin | Zhenyang Li | Jiang Zhong | Richard He Bai | Junnan Zhu
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
Kaiwen Wei | Kejun he | Xiaomian Kang | Jie Zhang | Ymyang | Li Jin | Zhenyang Li | Jiang Zhong | Richard He Bai | Junnan Zhu
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
Generative recommendation, which directly generates item identifiers, has emerged as a promising paradigm for recommendation systems. However, this left-to-right paradigm inherently biases the model towards local contexts, failing to capture deeper historical dependencies necessary for understanding complex user intents.To address this limitation, we propose Masked History Learning (MHL), a novel training framework that shifts the objective from simple next-step prediction to deep comprehension of history. MHL augments the standard autoregressive objective with an auxiliary task of reconstructing masked historical items, compelling the model to understand "why” an item path is formed from the user’s past behaviors, rather than just "what” item comes next.We introduce two key contributions to enhance this framework: (1) an entropy-guided masking policy that intelligently targets the most informative historical items for reconstruction, and (2) a curriculum learning scheduler that progressively transitions from history reconstruction to future prediction.Experiments on three public datasets show that our method significantly outperforms state-of-the-art generative models, highlighting that a comprehensive understanding of the past is crucial for accurately predicting a user’s future path. The code is available at https://github.com/CQU-MM-Intelligent-Lab/MHL.
CRAFT: A Unified Counterfactual Reasoning Framework for Tabular Question Answering and Fact Verification
Chenshuo Pan | Yu Zhao | Jie Zhang | Changzai Pan | Jiayi Liang | Yujie Mao | Zhenhe Wu | Shuangyong Song | Yongxiang Li | Zhongjiang He
Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)
Chenshuo Pan | Yu Zhao | Jie Zhang | Changzai Pan | Jiayi Liang | Yujie Mao | Zhenhe Wu | Shuangyong Song | Yongxiang Li | Zhongjiang He
Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)
Table reasoning remains challenging for large language models (LLMs), particularly in tasks that require multi-step inference over long and structured tables. Existing approaches predominantly rely on single-direction reasoning, which limits their ability to explore alternative hypotheses across tasks. In this work, we propose CRAFT, a unified Counterfactual Reasoning Framework that reformulates Tabular question answering and fact verification into a general bidirectional verification process. Our method explicitly constructs both declarative statements and their counterfactual variants. Evidence is then extracted from reasoning along both the original and counterfactual paths, and integrated via a weighted mechanism to arrive at the final answer. Experimental results show that our approach consistently surpasses representative baselines on table reasoning datasets such as WikiTQ and TabFact, achieving especially large improvements on complex question answering. Our framework also significantly mitigates performance gaps between different backbone LLMs. This indicates that counterfactual reasoning effectively overcomes the limitations of single-direction inference, guiding LLMs toward more discerning reasoning and establishing a more principled paradigm for structured reasoning tasks.
2025
T2R-BENCH: A Benchmark for Real World Table-to-Report Task
Jie Zhang | Changzai Pan | Sishi Xiong | Kaiwen Wei | Yu Zhao | Xiangyu Li | Jiaxin Peng | Xiaoyan Gu | Jian Yang | Wenhan Chang | Zhenhe Wu | Jiang Zhong | Shuangyong Song | Xuelong Li
Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing
Jie Zhang | Changzai Pan | Sishi Xiong | Kaiwen Wei | Yu Zhao | Xiangyu Li | Jiaxin Peng | Xiaoyan Gu | Jian Yang | Wenhan Chang | Zhenhe Wu | Jiang Zhong | Shuangyong Song | Xuelong Li
Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing
Extensive research has been conducted to explore the capabilities of large language models (LLMs) in table reasoning. However, the essential task of transforming tables information into reports remains a significant challenge for industrial applications. This task is plagued by two critical issues: 1) the complexity and diversity of tables lead to suboptimal reasoning outcomes; and 2) existing table benchmarks lack the capacity to adequately assess the practical application of this task. To fill this gap, we propose the table-to-report task and construct a bilingual benchmark named T2R-bench, where the key information flow from the tables to the reports for this task. The benchmark comprises 457 industrial tables, all derived from real-world scenarios and encompassing 19 industry domains as well as four types of industrial tables. Furthermore, we propose a novel evaluation criteria to fairly measure the quality of report generation. Expeimental results show that Deepseek-R1 only achieves the best performance with 62.71% overall score, indicating that LLMs still have room for improvement on T2R-bench.
Search
Co-authors
- Changzai Pan 2
- Shuangyong Song (宋双永) 2
- Kaiwen Wei 2
- Zhenhe Wu 2
- Yu Zhao 2
- Jiang Zhong 2
- Richard He Bai 1
- Wenhan Chang 1
- Jun Chen 1
- Shuangwu Chen 1
- Wang Dong 1
- Xiaoyan Gu 1
- Yu Guo 1
- Zhongjiang He (何忠江) 1
- Yingtian Hu 1
- Qiang Huang 1
- Jianyang 1
- Dong Jin 1
- Li Jin 1
- Xiaomian Kang 1
- Xiang Li 1
- Xiangyu Li 1
- Xuelong Li 1
- Yongxiang Li (李永翔) 1
- Zhengheng Li 1
- Zhenyang Li 1
- Ziman Li 1
- Jiayi Liang 1
- Wang Luo 1
- Wuyou Mao 1
- Yujie Mao 1
- Chenshuo Pan 1
- Jiaxin Peng 1
- Rui Qian 1
- Xiaobin Tan 1
- Haobo Wang 1
- Sishi Xiong 1
- Jian Yang 1
- Qi Yang 1
- Yuming Yang 1
- Shenghao Ye 1
- Wentao Ye 1
- Libin Zhan 1
- Enci Zhang 1
- Tao Zhang 1
- Quan Zheng 1
- Junnan Zhu 1
- Shujian Zhu 1
- Kejun he 1