cs.CL（2026-04-23）

📊 共 29 篇论文 | 🔗 5 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (22 🔗2) 支柱二：RL算法与架构 (RL & Architecture) (6 🔗3) 支柱一：机器人控制 (Robot Control) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (22 篇)

#题目一句话要点标签🔗⭐
1 A Multimodal Text- and Graph-Based Approach for Open-Domain Event Extraction from Documents 提出MODEE，结合图学习与LLM文本表示，解决开放域事件抽取中文档级推理难题。 large language model multimodal
2 Evaluation of Automatic Speech Recognition Using Generative Large Language Models 利用生成式大语言模型评估自动语音识别，提升语义相关性。 large language model
3 Revisiting Non-Verbatim Memorization in Large Language Models: The Role of Entity Surface Forms 提出RedirectQA数据集，研究LLM中实体表面形式对非字面记忆的影响 large language model
4 Process Supervision via Verbal Critique Improves Reasoning in Large Language Models 提出Verbal Process Supervision，通过外部语言反馈提升大语言模型推理能力 large language model
5 Unlocking the Power of Large Language Models for Multi-table Entity Matching 提出LLM4MEM框架，利用大语言模型解决多表实体匹配中的语义不一致和效率问题。 large language model ✅
6 EVENT5Ws: A Large Dataset for Open-Domain Event Extraction from Documents 提出EVENT5Ws：一个用于开放域文档事件抽取的超大型数据集。 large language model
7 From If-Statements to ML Pipelines: Revisiting Bias in Code-Generation 揭示代码生成中被低估的偏见：从条件语句到机器学习流水线 large language model
8 Language as a Latent Variable for Reasoning Optimization 提出polyGRPO，利用多语言作为隐变量优化LLM推理能力，提升跨任务泛化性。 chain-of-thought
9 Measuring Opinion Bias and Sycophancy via LLM-based Coercion 提出llm-bias-bench，通过多轮交互探测LLM在争议话题上的潜在偏见和谄媚行为。 large language model
10 Job Skill Extraction via LLM-Centric Multi-Module Framework 提出SRICL框架，解决LLM在职位技能抽取中边界漂移和幻觉问题 large language model
11 OptiVerse: A Comprehensive Benchmark towards Optimization Problem Solving OptiVerse：构建综合优化问题求解基准，揭示并缓解LLM在复杂优化任务中的建模瓶颈。 large language model
12 Reasoning Primitives in Hybrid and Non-Hybrid LLMs 研究混合与非混合LLM中的推理基元，揭示架构与推理增强对性能的影响 large language model
13 CARE: Counselor-Aligned Response Engine for Online Mental-Health Support CARE：针对在线心理健康支持的咨询师对齐回复引擎 large language model
14 When Bigger Isn't Better: A Comprehensive Fairness Evaluation of Political Bias in Multi-News Summarisation 针对多文档新闻摘要的政治偏见，提出综合评估框架并探索去偏见干预方法。 large language model
15 EngramaBench: Evaluating Long-Term Conversational Memory with Structured Graph Retrieval EngramaBench：通过结构化图检索评估长期对话记忆 large language model
16 Zero-Shot Detection of LLM-Generated Text via Implicit Reward Model 提出IRM：一种基于隐式奖励模型的LLM生成文本零样本检测方法 large language model
17 On Reasoning Behind Next Occupation Recommendation 提出基于推理的职业推荐方法，提升LLM在未来职业预测中的性能。 large language model ✅
18 Prefix Parsing is Just Parsing 提出前缀语法转换方法，将前缀解析高效规约到普通解析问题。 large language model
19 Lightweight Retrieval-Augmented Generation and Large Language Model-Based Modeling for Scalable Patient-Trial Matching 提出轻量级检索增强生成与大语言模型建模框架，用于可扩展的患者-试验匹配。 large language model multimodal
20 Shared Lexical Task Representations Explain Behavioral Variability In LLMs 词汇任务表征解释LLM行为变异性 large language model
21 Source-Modality Monitoring in Vision-Language Models 研究视觉-语言模型中源模态监控能力，揭示语法和语义信号的作用。 multimodal
22 When Cow Urine Cures Constipation on YouTube: Limits of LLMs in Detecting Culture-specific Health Misinformation 揭示LLM在检测文化特定健康虚假信息方面的局限性：以YouTube上的印度牛尿疗法为例 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (6 篇)

#题目一句话要点标签🔗⭐
23 Machine Behavior in Relational Moral Dilemmas: Moral Rightness, Predicted Human Behavior, and Model Decisions 基于关系型道德困境，揭示LLM决策与人类社会规范的偏差 world model world models large language model
24 Explainable Disentangled Representation Learning for Generalizable Authorship Attribution in the Era of Generative AI 提出EAVAE，通过可解释的解耦表示学习提升生成AI时代作者身份归属的泛化性 representation learning contrastive learning ✅
25 SemEval-2026 Task 4: Narrative Story Similarity and Narrative Representation Learning SemEval-2026任务4：提出一种新的叙事相似度定义，并构建数据集用于叙事表示学习。 representation learning
26 When Agents Look the Same: Quantifying Distillation-Induced Similarity in Tool-Use Behaviors 提出RPS和AGS指标，量化工具使用行为中蒸馏诱导的相似性，揭示Agent同质化现象 distillation ✅
27 Outcome Rewards Do Not Guarantee Verifiable or Causally Important Reasoning 揭示基于结果奖励的强化学习在推理中存在的因果性与可验证性问题 reinforcement learning chain-of-thought
28 Incentivizing Neuro-symbolic Language-based Reasoning in VLMs via Reinforcement Learning 通过强化学习激励视觉语言模型中的神经符号语言推理 reinforcement learning ✅

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
29 Cross-Entropy Is Load-Bearing: A Pre-Registered Scope Test of the K-Way Energy Probe on Bidirectional Predictive Coding 研究交叉熵损失在双向预测编码网络能量探针中的作用 manipulation

⬅️ 返回 cs.CL 首页 · 🏠 返回主页