Knowledge Extraction Tool for System Model Generation in an LLM-Enhanced Spacecraft Design Assistant

The potential of Knowledge graphs (KGs) to provide sophisticated querying, intelligent data management, and effective knowledge representation is driving their use in a large number of domains. This paper presents a framework for building Knowledge Graphs (KGs) using Large Language Models (LLMs) and demonstrates how these KGs can be used to build space mission system models. The study introduces a knowledge extraction tool that uses LLMs to process unstructured text and turn it into structured knowledge graphs. These KGs—which are made up of entities and relationships —are then used to map data to OPM (Object Process Methodology) models. The OPM model’s elements and hierarchical structure is used as an ontology to label the entities that are identified during the KG construction. This study assesses the tool’s ability to accurately map the retrieved data onto the OPM model in addition to producing KGs. Although the LLM worked well for extracting entities and relationships especially for producing high-level system components—it had trouble labeling lower-level, domain-specific entities in the Knowledge Graph. Despite these limitations, this framework provides a promising initial solution and establishes the groundwork for future research and development targeted at obtaining a more thorough automatic population of OPM models.     «

The potential of Knowledge graphs (KGs) to provide sophisticated querying, intelligent data management, and effective knowledge representation is driving their use in a large number of domains. This paper presents a framework for building Knowledge Graphs (KGs) using Large Language Models (LLMs) and demonstrates how these KGs can be used to build space mission system models. The study introduces a knowledge extraction tool that uses LLMs to process unstructured text and turn it into structured k...     »