?
Проектирование инструментов визуализации данных на основе интеграции возможностей предметно-ориентированного моделирования и генеративного искусственного интеллекта
Abstract. Data visualization tools are key analytics tools that make it easier to identify dependencies, trends, and patterns. These tools are used by a broad range of specialists (analysts, scientists, business leaders, managers, teachers, and specialists in other fields where accurate and understandable presentation of information is critically important to improve the effectiveness of data analysis). Existing visualization tools limit customization options and offer only basic parameter settings, which do not always meet the needs of users, as some scenarios require more in-depth modifications, including the ability to create custom chart types. Such deep customization often requires the use of programming languages, which creates a barrier for users without technical skills. A new technique for developing custom data visualizations is proposed, which combines the advantages of three approaches: a language-oriented approach to developing effective visualization models using automation tools for developing languages and generating code for implementing visualizations; existing libraries addition with custom data visualization packages; automatic script generation for implementing custom visualizations based on prompts to language models, capabilities of generative artificial intelligence. The article describes the architecture of data visualization system and the multifaceted ontology that is the core of the system. Examples of using the developed research prototype of tools for creating custom visualizations are given, showing the practical significance of the proposed approach. The results obtained are evaluated and the prospects for the development of the tools described are shown.