Visualization of the semantics of text descriptions presented in various formats

• Дата: 2025
• Автор: Minenko, V.D.
• Формат: Стаття
• Мова: Ukrainian
• Опубліковано: PROBLEMS IN PROGRAMMING 2025
• Теми: visualization of semantics; semantically meaningful elements; generative artificial intelligence; natural language processing; text analysis methods; tokenization; lemming; segmentation; AI generation model; generative adversarial network; machine learnin; UDC 004.94
• Онлайн доступ: https://pp.isofts.kiev.ua/index.php/ojs1/article/view/768
• Назва журналу: Problems in programming

 This study is aimed at solving the problem of identifying semantics from arbitrary texts presented in various formats and further visualizing it using modern tools of generative artificial intelligence. The rapid development of artificial intelligence technologies provides fundamentally new opportunities for solving both text analysis tasks and content generation - visualizations (in the form of images or videos). As a result, we can talk about a different, modern level of solving applied problems using similar functionality. The field of generative artificial intelligence is still quite young and contains many unsolved problems. The generated visualization is characterized not only by the technical quality of the image or video, but also by the adequacy of the presentation of the semantics of the input text description, which usually directly depends not only on the possibility of the selected AI tool, but also on the structure and content of the input text prompt. This article describes the algorithm to form a chain of solving the given task, from the criteria for choosing tools of developments and identifying problems that need improvement or resolving, to determining the scheme of a composite solution. The method created within the framework of the proposed study has certain limitations, namely: it does not support multilingual content and does not cover the processing of dialects, slangs, automatic detection of the language of the text.Problems in programming 2025; 1: 94-109