AUTOMATIC DOCUMENTATION AND MATHEMATICAL LINGUISTICS最新文献

Generative Large Language Models for Word Sense Disambiguation in Russian 俄语词义消歧的生成式大型语言模型

IF 0.4

AUTOMATIC DOCUMENTATION AND MATHEMATICAL LINGUISTICS Pub Date : 2026-10-13 DOI: 10.3103/S0005105526700731

P. A. Gousyatskaya, N. V. Loukachevitch

{"title":"Generative Large Language Models for Word Sense Disambiguation in Russian","authors":"P. A. Gousyatskaya, N. V. Loukachevitch","doi":"10.3103/S0005105526700731","DOIUrl":"10.3103/S0005105526700731","url":null,"abstract":"This paper is dedicated to experiments in Word Sense Disambiguation (WSD) on Russian-language material using generative (decoder-only) models of small and medium size. While the direct application of generative models is not an optimal approach for this task, such models demonstrate potential as primary semantic annotators of raw unstructured data. The application of generative models to semantic annotation may help overcome the bottleneck of insufficient annotated data for training encoder-based models. As previous studies have shown, flagship English-language and multilingual generative models can achieve accuracy above 90% on this task, while smaller models typically reach 80%+. The present study aims to determine whether a similar task can be solved on Russian-language material using small- and medium-size Russian-adapted models (up to 32B) that do not require substantial computational resources for their use. The disambiguation experiments were conducted both in the basic setting (one-shot prompting) and in various modifications (context enrichment with dictionary information—hypernyms, hyponyms, domain labels, etc.; analysis of broad and narrow context windows of the ambiguous lexeme; ensemble approaches in which one model validates and corrects the predictions of another). The experiments are based on the Russian corpus RuSemCor, annotated with the categories of the RuWordNet semantic database. According to the results of the experiments, the models proved suitable for the task at hand: all the models perform above the random-prediction level, and the most powerful ones achieve 80% accuracy, which is comparable to the results of English-language models of the same size. The broad context of the ambiguous lexeme proved more informative for the models. Approaches with input-data enrichment and ensemble methods yielded a significant gain in quality.","PeriodicalId":42995,"journal":{"name":"AUTOMATIC DOCUMENTATION AND MATHEMATICAL LINGUISTICS","volume":"60 4","pages":"S423 - S428"},"PeriodicalIF":0.4,"publicationDate":"2026-10-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149176430","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Intelligence, Productive Thinking, and Artificial Intelligence 智能、创造性思维和人工智能

IF 0.4

AUTOMATIC DOCUMENTATION AND MATHEMATICAL LINGUISTICS Pub Date : 2026-10-13 DOI: 10.3103/S0005105526700949

V. K. Finn

引用次数: 0

Problem of Building Synthetic Psychological Data: Experience in Modeling Reactions to Frustration 建立综合心理数据的问题：挫折反应建模的经验

IF 0.4

AUTOMATIC DOCUMENTATION AND MATHEMATICAL LINGUISTICS Pub Date : 2026-10-13 DOI: 10.3103/S0005105526700792

A. A. Chuganskaya, D. A. Kireev, I. V. Smirnov, O. G. Grigoriev

{"title":"Problem of Building Synthetic Psychological Data: Experience in Modeling Reactions to Frustration","authors":"A. A. Chuganskaya, D. A. Kireev, I. V. Smirnov, O. G. Grigoriev","doi":"10.3103/S0005105526700792","DOIUrl":"10.3103/S0005105526700792","url":null,"abstract":"The issue of generating synthetic data for psychological research remains relevant and complex. Problems of the confidentiality, reliability, validity, and accuracy of conclusions are unevenly represented in different areas of psychology and are actually related to the use of synthetic data in related sciences, such as medicine, sociology, history, political science, and economics. The study of various psychological phenomena in large social groups is associated with the analysis of complex and difficult-to-formalize constructs. Synthetic data refers to artificially generated data based on algorithms and modeling. S. Rosenzweig’s classification of types of reactions to frustration was chosen as the basis for this study. When analyzing online discourse, there is a problem of the low number of certain types. This is especially true for the class of impunitive reactions. The paper analyzes the possibility of the creation of text corpora using synthetic data of reactions to frustration that is generated by large language models. In the experiments, the experts created prompts and generated examples of impulsive reactions using four large language models, with 10 examples of each type of reaction. They then evaluated the contextual validity and quality of the generated responses. The results allowed them to identify the weaknesses in generating texts with complex psychological phenomena for training neural network models.","PeriodicalId":42995,"journal":{"name":"AUTOMATIC DOCUMENTATION AND MATHEMATICAL LINGUISTICS","volume":"60 4","pages":"S468 - S473"},"PeriodicalIF":0.4,"publicationDate":"2026-10-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149176384","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Application of Quantized Algorithms for Language Model Adaptation in the Task of Verifying the Solution Process of Quadratic Equations 量化语言模型自适应算法在二次方程解过程验证任务中的应用

IF 0.4

AUTOMATIC DOCUMENTATION AND MATHEMATICAL LINGUISTICS Pub Date : 2026-10-13 DOI: 10.3103/S0005105526700895

A. N. Khaybullin, D. N. Tumakov

{"title":"Application of Quantized Algorithms for Language Model Adaptation in the Task of Verifying the Solution Process of Quadratic Equations","authors":"A. N. Khaybullin, D. N. Tumakov","doi":"10.3103/S0005105526700895","DOIUrl":"10.3103/S0005105526700895","url":null,"abstract":"The work investigates quantized approaches to language model adaptation for the task of automatic step-by-step verification of the correctness of quadratic equation solutions. The study examines the effectiveness of Parameter-Efficient Fine-Tuning (PEFT) methods when adapting the DeepSeek-R1-Distill-Qwen-1.5B and InternLM2-Math-Plus-1.8B language models to build a mathematical verifier (Process-supervised Reward Model, PRM). Experiments are conducted on a synthetic dataset of quadratic equations augmented with negative sampling to simulate learner errors. A comparative evaluation of standard (LoRA, DoRA, rsLoRA) and quantized (QLoRA, QDoRA, LoftQ) fine-tuning algorithms is performed. Zero-shot transfer generalization is additionally assessed on a structurally distinct dataset of linear equations. Results show that quantization resolves numerical stability issues for non-standard architectures (InternLM2) while achieving quality comparable to standard methods. For the DeepSeek-R1 model, QLoRA, LoftQ, and QDoRA reached accuracies of 97.77, 98, and 98% respectively, only marginally below the standard LoRA (98.67%). Similarly, for the non-standard InternLM2 architecture, QLoRA achieved 92.67% accuracy (vs. 93% for baseline LoRA). However, full-precision algorithms (LoRA) tend to preserve richer representations of learned patterns, providing better knowledge transfer for Reasoning-class models (DeepSeek-R1 accuracy 66.8 vs. 61.4% for QLoRA on unseen data).","PeriodicalId":42995,"journal":{"name":"AUTOMATIC DOCUMENTATION AND MATHEMATICAL LINGUISTICS","volume":"60 4","pages":"S538 - S547"},"PeriodicalIF":0.4,"publicationDate":"2026-10-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149176478","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Determination of the Effective Mechanical Characteristics of a Nonlinear Composite Material with Spherical Fillers 球面填料非线性复合材料有效力学特性的测定

IF 0.4

AUTOMATIC DOCUMENTATION AND MATHEMATICAL LINGUISTICS Pub Date : 2026-10-13 DOI: 10.3103/S0005105526700834

N. R. Battalov, I. R. Garifullin, L. U. Sultanov, L. R. Fakhrutdinov

引用次数: 0

A Systematic Approach to the Analysis of Threat Models for Technical Information Leak Channels 技术信息泄露渠道威胁模型的系统分析方法

IF 0.4

AUTOMATIC DOCUMENTATION AND MATHEMATICAL LINGUISTICS Pub Date : 2026-10-13 DOI: 10.3103/S0005105526700925

K. V. Pitelinsky, V. V. Britvina, K. A. Gruzdov, E. O. Bobrova, S. O. Makovey

引用次数: 0

Adaptive RAG Architecture for Intelligent Search in a Corpus of Documents of Educational Institutions 面向高校文献语料库智能检索的自适应RAG体系结构

IF 0.4

AUTOMATIC DOCUMENTATION AND MATHEMATICAL LINGUISTICS Pub Date : 2026-10-13 DOI: 10.3103/S0005105526700858

A. D. Budrevich, M. M. Abramskiy, I. A. Valishin

引用次数: 0

LLM Confidence in Constructing Semantic Classifications of Arguments 构建论点语义分类的LLM信心

IF 0.4

AUTOMATIC DOCUMENTATION AND MATHEMATICAL LINGUISTICS Pub Date : 2026-10-13 DOI: 10.3103/S0005105526700755

D. S. Larionov, E. N. Nikitina, I. V. Smirnov

{"title":"LLM Confidence in Constructing Semantic Classifications of Arguments","authors":"D. S. Larionov, E. N. Nikitina, I. V. Smirnov","doi":"10.3103/S0005105526700755","DOIUrl":"10.3103/S0005105526700755","url":null,"abstract":"This article addresses the problem of quantifying the confidence of large language models (LLMs) in the automatic semantic classification of arguments with emotive predicates. Using Russian-language social media posts, we analyze verbs of fear (to scare, to be afraid, etc.) and emotional attitude (to like, to love) with the semantic roles of experiencer, causator, and object. The study aims to compare self-assessed confidence of LLM Claude Sonnet 4.5 with expert assessments of the model’s reasoning texts in argument classification in the healthcare domain. The experiment utilized a stratified set of 300 examples using chain-of-thoughts reasoning in Russian and a four-step confidence scale. The results showed a moderate Spearman’s correlation between the expert and model assessments. A statistically significant relationship was found only between self-assessment of the model and the actual classification accuracy, while expert assessment of the linguistic characteristics of reasoning was unrelated to accuracy. It was concluded that explicit LLM reasoning is not directly related to its self-assessment of confidence and is separate from the decision-making process; reasoning may be an important functional part of the user interface, but not that of the research.","PeriodicalId":42995,"journal":{"name":"AUTOMATIC DOCUMENTATION AND MATHEMATICAL LINGUISTICS","volume":"60 4","pages":"S438 - S444"},"PeriodicalIF":0.4,"publicationDate":"2026-10-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149176640","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Specifics of Assessing Human Professional Cognitive Abilities when Justifying Quality Requirements for Artificial Intelligence Systems 在证明人工智能系统质量要求时评估人类专业认知能力的细节

IF 0.4

AUTOMATIC DOCUMENTATION AND MATHEMATICAL LINGUISTICS Pub Date : 2026-10-13 DOI: 10.3103/S0005105526700950

S. V. Garbuk

引用次数: 0

Extracting New Translation Variants of Polysemous Words from Parallel Texts 从平行文本中提取多义词翻译新变体

IF 0.4

AUTOMATIC DOCUMENTATION AND MATHEMATICAL LINGUISTICS Pub Date : 2026-10-13 DOI: 10.3103/S0005105526700962

I. M. Zatsman, S. D. Ignatova

引用次数: 0