Automatic generation of SysML requirement models based on Chinese natural language requirements

Abstract

In the face of increasingly complex systems engineering R&D, Model-Based Systems Engineering (MBSE) is gradually replacing Traditional Systems Engineering, producing a significant shift in modelling language, modelling thinking and modelling tools. And SysML, as a common modelling language for MBSE practitioners, has many irreplaceable advantages. Requirements diagrams are the primary medium of SysML for conveying traceability between requirements and from requirements to structure and behaviour in a system model. Today, how to automate the construction of SysML requirements models is a significant challenge. This paper proposes an automatic SysML model generation method based on Chinese free-text requirements using natural language processing. Firstly, perform word segmentation and POS Tagging on the requirement text to construct a domain lexicon, extract effective requirements by combining the domain lexicon and regular expressions, and then mark the effective requirements with semantic roles, and determine the requirements elements through established rules, which are ultimately converted into SysML requirements diagrams.