Rapid Generation of SysML Models using NLP/LLM

The proposed project is to demonstrate the feasibility of building a NLP/LLM based application that accelerates the generation of SysML models of complex space systems by smartly extracting information from textual descriptions, and transforming these into SysML model elements. One of the primary challenges in creating SysML models from textual technical documents is the diversity of document types, writing styles, and languages. NLP and LLMs can help address these challenges by employing advanced techniques for text preprocessing, entity recognition, and semantic analysis. NLP, a subfield of artificial intelligence, focuses on the interaction between computers and human language. It equips machines with the ability to understand, interpret, and generate human language, making it a valuable technology for processing unstructured textual information. Accelerating the development of SysML models is critical in order to support future space missions by making model development cost effective. NASA centers are using Model Based System Engineering (MBSE) and creating Systems Modeling Language (SysML) representations of their systems, following best Systems Engineering practices. These models are difficult and expensive to develop. NASA MBSE based system engineering development could be dramatically more efficient through the use of a toolset that can automatically generate the various SysML models from textual descriptions such as requirements documents, concept of operations, Design Reference Mission (DRM) and other technical descriptions. Such a toolset can be applied to a wide range of projects within NASA centers and could be adapted to other agencies and industries that are using the MBSE approach.