GRAMMAR-BASED METHOD FOR TRANSFORMING FORMALIZED REQUIREMENTS INTO EXTENDED UML MODELS

Abstract

This article addresses the relevant task of transforming formalized requirements into extended UML models, which is of great importance for increasing the efficiency of designing intelligent and critical systems. The study focuses on combining a grammatical approach with UML profile mechanisms, which makes it possible to ensure automated consistency checking of requirements, representation of uncertainty and probabilistic properties, and improvement of the accuracy and adaptability of the created models. The relevance of the topic is determined by the increasing complexity of modern information systems, where requirements often remain vague, contradictory, or changing, which complicates their formalization and further representation in traditional UML.A methodology is proposed based on a two-level grammar: attribute grammar is used for requirements reconciliation and for constructing the their consistent set, while transformation grammar defines the mapping of these requirements into UML elements according to their attributes (confidence, likelihood, context, priority, resources). The article provides formal transformation rules for functional, non-functional, data, and scenario requirements, which are implemented through the creation of use cases, classes, constraints, sequence diagrams, and state diagrams. Special attention is paid to the mechanism of assigning stereotypes that preserve the semantics of the original specifications, as well as to tracing, which ensures transparent links between requirements and UML model elements.The practical applicability of the method is demonstrated using the example of an intelligent medical patient monitoring system. The study revealed several advantages of the developed approach: reduction of the time required for updating models, improvement of the level of consistency between requirements and their model representations, and the ability to represent fuzzy and probabilistic properties. A comparison with classical UML without a profile showed that the proposed method provides broader opportunities for modeling adaptive systems in environments with a high level of uncertainty.At the same time, certain limitations were identified: the need to customize grammar for a specific application domain and the requirement for integration with CASE tools to automate transformations.Further research involves several directions for the development of the methodology: automated generation of UML diagrams in modern CASE tools; as well as integration of the approach with SysML v2 as a promising system engineering standard.