Papers by Qingcheng Zeng
Verified Critical Step Optimization for LLM Agents (2026.findings-acl)
Copied to clipboard
| Challenge: | Critical Step Optimization (CSO) focuses preference learning on verified critical steps where alternative actions demonstrably flip task outcomes from failure to success. |
| Approach: | They propose a method which focuses preference learning on verified critical steps where alternative actions demonstrably flip task outcomes from failure to success. |
| Outcome: | The proposed method outperforms the existing methods on GAIA-Text-103 and XBench-DeepSearch while requiring supervision at only 16% of trajectory steps. |
A Survey in Automatic Irony Processing: Linguistic, Cognitive, and Multi-X Perspectives (2022.coling-1)
Copied to clipboard
| Challenge: | figurative language research has focused on sarcasm and irony, but there is still a gap in the field. |
| Approach: | They propose to review computational irony, cognitive science, and neural models of irony processing . they aim to encourage a balanced and equal research environment in figurative languages . |
| Outcome: | The proposed multi-X irony processing perspectives will provide an overview of computational irony, insights from linguisic theory and cognitive science, and interactions with downstream NLP tasks. |
The Confidence Dichotomy: Analyzing and Mitigating Miscalibration in Tool-Use Agents (2026.acl-long)
Copied to clipboard
| Challenge: | a fundamental pillar of trustworthiness is calibration, which refers to an agent’s ability to express confidence that reliably reflects its actual performance. |
| Approach: | They propose a reinforcement learning framework that jointly optimizes task accuracy and calibration, supported by a holistic benchmark of reward designs. |
| Outcome: | The proposed framework improves calibration across tool types and shows that trained agents achieve superior calibration and exhibit robust generalization from local training environments to noisy web settings and to distinct domains such as mathematical reasoning. |
Good Intentions Beyond ACL: Who Does NLP for Social Good, and Where? (2025.emnlp-main)
Copied to clipboard
| Challenge: | 20% of all papers in the ACL Anthology address social good issues . authors are more likely to do work addressing social good concerns when publishing in venues outside of ACL. |
| Approach: | They use author- and venue-level perspectives to map the landscape of NLP4SG . they find authors are more likely to do work addressing social good concerns outside of ACL . |
| Outcome: | The study analyzes the literature on NLP4SG and its impact on the ACL community . 20% of all papers in the anthology address social good issues, the study finds . |
Code-Switching Information Retrieval: Benchmarks, Analysis, and the Limits of Current Retrievers (2026.findings-acl)
Copied to clipboard
Qingcheng Zeng, Yuheng Lu, Zeqi Zhou, Heli Qi, Puxuan Yu, Fuheng Zhao, Hitomi Yanaka, Weihao Xuan, Naoto Yokoya
| Challenge: | a new study examines the performance of code-switching IR in monolingual contexts . code-witching is a pervasive linguistic phenomenon in global communication . |
| Approach: | They propose a benchmark to evaluate code-switching IR in monolingual contexts . they propose CS-MTEB, which measures performance declines of up to 27% . |
| Outcome: | The proposed benchmark shows that code-switching performance is degraded by 27% . the proposed benchmark is based on a dataset of mixed-language queries . |
On the Effect of Hyperparameters in Language Modeling for Computational Linguistics (2026.acl-long)
Copied to clipboard
| Challenge: | Training language models and examining their linguistic behaviors is a common protocol in computational linguistics for studying linguistic phenomena and modeling human language processing. |
| Approach: | They replicate three prior studies with hyperparameters varied within a practical range and show that modest hyperparametric changes can alter qualitative conclusions about models’ linguistic abilities. |
| Outcome: | The results show that hyperparameter changes can alter qualitative conclusions and reverse the ranking of models. |
Thinking Out Loud: Do Reasoning Models Know When They’re Right? (2025.emnlp-main)
Copied to clipboard
| Challenge: | Large reasoning models (LRMs) have recently demonstrated impressive capabilities in complex reasoning tasks by leveraging increased test-time computation and exhibiting behaviors reminiscent of human-like self-reflection. |
| Approach: | They analyze verbalized confidence, how models articulate their certainty, as a lens into the nature of self-reflection in large reasoning models. |
| Outcome: | The proposed model exhibits human-like self-reflection in reasoning tasks, but how this ability interacts with other model behaviors remains underexplored. |
The Pragmatic Mind of Machines: Tracing the Emergence of Pragmatic Competence in Large Language Models (2026.eacl-long)
Copied to clipboard
| Challenge: | Current large language models (LLMs) have demonstrated emerging capabilities in social intelligence tasks, including implicature resolution and theory-of-mind reasoning. |
| Approach: | They introduce a dataset grounded in the pragmatic concept of alternatives to evaluate whether large language models can accurately infer nuanced speaker intentions. |
| Outcome: | The proposed model can infer nuanced speaker intentions by inferring the speaker’s intended meaning and explaining when and why a speaker would choose one utterance over its alternative. |
Adaptive Axes: A Pipeline for In-domain Social Stereotype Analysis (2024.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods to quantify social stereotypes have struggled to capture the variability in stereotypes across conceptual domains for the same social group. |
| Approach: | They propose to use text embedding models and adaptive semantic axes to recover stereotypes from contextual representations by using large language models. |
| Outcome: | The proposed pipeline surpasses token-based methods in capturing in-domain framing and tracks stereotypes along domain-specific semantic axes for in- domain texts. |
Exploring Concept Depth: How Large Language Models Acquire Knowledge and Concept at Different Layers? (2025.coling-main)
Copied to clipboard
Mingyu Jin, Qinkai Yu, Jingyuan Huang, Qingcheng Zeng, Zhenting Wang, Wenyue Hua, Haiyan Zhao, Kai Mei, Yanda Meng, Kaize Ding, Fan Yang, Mengnan Du, Yongfeng Zhang
| Challenge: | Large language models have shown remarkable performances across a wide range of tasks, but mechanisms by which they encode tasks of varying complexity remain poorly understood. |
| Approach: | They propose to explore the possibility that LLMs process concepts in different layers . they propose to categorize concepts based on their level of abstraction . |
| Outcome: | The proposed model can process complex concepts in shallow layers, the authors show . the proposed model could be used to prob complex tasks in shallow ones . |
Seeing is Believing, but How Much? A Comprehensive Analysis of Verbalized Calibration in Vision-Language Models (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing approaches to quantify uncertainty are limited in vision-language models . however, current models display notable miscalibration across diverse tasks and settings . |
| Approach: | They evaluate verbalized confidence in vision-language models using visual reasoning . they propose a prompting strategy that improves confidence alignment in multimodal settings . |
| Outcome: | The proposed method improves confidence alignment across multimodal settings. |
Leveraging Human Production-Interpretation Asymmetries to Test LLM Cognitive Plausibility (2025.acl-short)
Copied to clipboard
| Challenge: | Existing research on the linguistic capabilities of large language models has focused on their performance in language interpretation. |
| Approach: | They examine whether large language models (LLMs) process language similarly to humans . they use an empirically documented asymmetry between production and interpretation in humans a testbed . |
| Outcome: | The proposed model can replicate human-like distinctions between production and interpretation. |
MMLU-ProX: A Multilingual Benchmark for Advanced Large Language Model Evaluation (2025.emnlp-main)
Copied to clipboard
Weihao Xuan, Rui Yang, Heli Qi, Qingcheng Zeng, Yunze Xiao, Aosong Feng, Dairui Liu, Yun Xing, Junjue Wang, Fan Gao, Jinghui Lu, Yuang Jiang, Huitao Li, Xin Li, Kunyu Yu, Ruihai Dong, Shangding Gu, Yuekang Li, Xiaofei Xie, Felix Juefei-Xu, Foutse Khomh, Osamu Yoshie, Qingyu Chen, Douglas Teodoro, Nan Liu, Randy Goebel, Lei Ma, Edison Marrese-Taylor, Shijian Lu, Yusuke Iwasawa, Yutaka Matsuo, Irene Li
| Challenge: | Existing large language model evaluation benchmarks focus on English, while current multilingual tasks lack parallel questions that specifically assess cross-lingual reasoning abilities. |
| Approach: | They propose a comprehensive benchmark covering 29 languages, built on an English benchmark. |
| Outcome: | The MMLU-ProX is a comprehensive benchmark covering 29 languages, built on an English benchmark. |
DeepSieve: Information Sieving via LLM-as-a-Knowledge-Router (2026.findings-eacl)
Copied to clipboard
Minghao Guo, Qingcheng Zeng, Xujiang Zhao, Yanchi Liu, Wenchao Yu, Mengnan Du, Haifeng Chen, Wei Cheng
| Challenge: | Existing RAG methods lack fine-grained control over query and source sides, resulting in noisy retrieval and shallow reasoning. |
| Approach: | They propose an agentic RAG framework that integrates information sieving via LLM-as-a-knowledge-router. |
| Outcome: | Experiments on multi-hop QA tasks across heterogeneous sources demonstrate improved reasoning depth, retrieval precision, and interpretability over conventional approaches. |
Evaluating Large Language Models on Wikipedia-Style Survey Generation (2024.findings-acl)
Copied to clipboard
Fan Gao, Hang Jiang, Rui Yang, Qingcheng Zeng, Jinghui Lu, Moritz Blum, Tianwei She, Yuang Jiang, Irene Li
| Challenge: | Recent studies have shown that large language models can perform well in general tasks, but their effectiveness and limitations in domainspecific tasks remain unclear. |
| Approach: | They examine the proficiency of Large Language Models (LLMs) in generating succinct survey articles specific to the niche field of NLP in computer science. |
| Outcome: | The LLMs perform better in generating succinct survey articles specific to the niche field of NLP in computer science, compared to human-authored surveys, but they exhibit bias in evaluation. |
Large Language Models Are Partially Primed in Pronoun Interpretation (2023.findings-acl)
Copied to clipboard
| Challenge: | Existing studies suggest large language models acquire rich linguistic representations, but little is known about whether they adapt to linguistic biases in a human-like way. |
| Approach: | They examine whether large language models display human-like referential biases using stimuli and procedures from real psycholinguistic experiments. |
| Outcome: | The proposed models display human-like referential biases when exposed to referential patterns in the local context. |
How to Improve LLMs’ Performance on Specific Languages: A Perspective on LLM-Derived Language Similarity (2026.acl-long)
Copied to clipboard
| Challenge: | Large language models (LLMs) exhibit uneven performance across languages. |
| Approach: | They propose to use a framework to quantify the similarity within each language pair through both the lenses of language-specific performance patterns and cross-lingual transferability. |
| Outcome: | The proposed approach outperforms traditional linguistic typology and cross-lingual transferability measures on multilingual LLMs. |