Automatic Generation of Scientific Articles Abstracts Based on Large Language Models

Informatics and Automation, Год журнала: 2025, Номер 24(1), С. 275 - 301

Опубликована: Янв. 20, 2025

The concept of automation the process annotation scientific materials (Russian-language articles) is proposed and its practical implementation carried out by means machine learning technologies, additional training large language models. relevance correct rational compilation annotations indicated, problems related to establishing a balance between time-consuming ensuring compliance with key requirements for are highlighted. basics presented in family standards on information, librarianship, publishing analyzed, classification their content functionality given. essence process, typical structure research object schematically. issue integration digital technologies into special attention paid advantages introducing artificial intelligence technology. toolkit used generate text natural processing applications briefly described. Its shortcomings solving problem posed this article noted. part substantiates choice model solve conditional generation. existing pre-trained models analyzed and, considering statement limitations computing resources, ruT5-base selected. A description dataset given, including articles from journals included list peer-reviewed publications which main results dissertations degrees candidate doctor science should be published. data labeling technique based operation tokenizer characterized, numerical characteristics distributions parameters pipeline graphically tables. ROUGE quality metric evaluate model, expert assessment method, grammar logic as basic criteria, results. automatic generation comparable real texts meets information content, compactness. may interest an audience scientists researchers seeking optimize activities terms integrating digitalization tools writing articles, well specialists involved

Язык: Английский

A Comprehensive Metadata Framework for Preservation and Accessibility of Digital News and Educational Resource Management

SN Computer Science, Год журнала: 2025, Номер 6(5)

Опубликована: Май 2, 2025

Язык: Английский