Princeton University（普林斯顿大学）；Beijing Institute of Big Data Research（北京大数据研究院）
Weinan E, Yajun Zhou
备注：
Main text (12 pages, 7 figures); Software Manual (ii+262 pages, 16 figures, 12 tables, available as two ancillary files). Revised according to reviewers' comments
We characterize the meaning of words with language-independent numerical fingerprints, through a mathematical analysis of recurring patterns in texts. Approximating texts by Markov processes on a long-range time scale, we are able to extract topics, discover synonyms, and sketch semantic fields from a particular document of moderate length, without consulting external knowledge-base or thesaurus. Our Markov semantic model allows us to represent each topical concept by a low-dimensional vector, interpretable as algebraic invariants in succinct statistical operations on the document, targeting local environments of individual words. These language-independent semantic representations enable a robot reader to both understand short texts in a given language (automated question-answering) and match medium-length texts across different languages (automated word translation). Our semantic fingerprints quantify local meaning of words in 14 representative languages across 5 major language families, suggesting a universal and cost-effective mechanism by which human languages are processed at the semantic level. Our protocols and source codes are publicly available on https://github.com/yajun-zhou/linguae-naturalis-principia-mathematica
Universal Information Extraction as Unified Semantic Matching
Baidu Inc.（百度公司）；Chinese Information Processing Laboratory（中文信息处理实验室）；State Key Laboratory of Computer Science（计算机科学国家重点实验室）；Institute of Software, Chinese Academy of Sciences（中国科学院软件研究所）
The Universal Decompositional Semantics Dataset and Decomp Toolkit
University of Rochester（罗切斯特大学）；Johns Hopkins University（约翰斯·霍普金斯大学）；University of Maryland Baltimore County（马里兰大学巴尔的摩县分校）；Allen Institute for Artificial Intelligence（艾伦人工智能研究所）；University of Texas at Austin（德克萨斯大学奥斯汀分校）
COME: Dual Structure-Semantic Learning with Collaborative MoE for Universal Lesion Detection Across Heterogeneous Ultrasound Datasets
College of Artificial Intelligence, Nanjing University of Aeronautics and Astronautics（人工智能学院，南京航空航天大学）；ByteDance Inc.（字节跳动公司）；Tsinghua University（清华大学）；Shanghai Jiaotong University（上海交通大学）
On the universal structure of human lexical semantics
Institute for New Economic Thinking at the Oxford Martin School（牛津马丁学院新经济思维研究所）；University of Oxford（牛津大学）；Santa Fe Institute（圣塔菲研究所）；University of New Mexico（新墨西哥大学）；Ronin Institute（浪人研究所）；Los Alamos National Laboratory（洛斯阿拉莫斯国家实验室）
Efficient Universal Goal Hijacking with Semantics-guided Prompt Organization
Nanyang Technological University, Singapore（南洋理工大学）；CFAR and IHPC, Agency for Science, Technology and Research (A*STAR), Singapore（CFAR和IHPC，新加坡科技研究局）；New York University, USA（纽约大学）；East China Normal University, China（华东师范大学）
Autoformalize Mathematical Statements by Symbolic Equivalence and Semantic Consistency
State Key Lab of Novel Software Technology, Nanjing University（南京大学现代软件技术国家重点实验室）；Peking University（北京大学）；University of Toronto（多伦多大学）；Microsoft Research Asia（微软亚洲研究院）
针对 LLM 数学命题自动形式化中 pass@1 与 pass@k 的差距，提出结合符号等价性和语义一致性从 k 个候选中选择最佳结果的框架，在 MATH 与 miniF2F 上显著提升准确率。
A survey of cross-lingual features for zero-shot cross-lingual semantic parsing
Georgia Institute of Technology（佐治亚理工学院）；Samsung AI Center（三星人工智能中心）；University of Edinburgh（爱丁堡大学）
利用语料库来训练英语语义分析器，已经在该领域取得了重大进展。不幸的是，对于英语以外的语言，注释是稀缺的，所以是发达的解析器。然后我们会问：一个用英语训练的解析器可以应用于它没有被训练过的语言吗？为了回答这个问题，我们探索zero-shot跨语言语义解析，其中我们训练可用的从粗到细的语义解析器（Liu et al.，2018）使用跨语言词嵌入和英语中的通用依赖关系，并在意大利语，德语和荷兰语上进行测试。并行意义库（一个多语言语义图库）的结果表明，当与其他词汇特征结合使用时，通用依赖特征显着提高了性能，但在对输入进行编码时，直接对UD结构进行建模则不会。
