Transforming natural language requirements for aerospace systems to formalism using large language models

A processor-implemented method for using a large language model (LLM) to convert natural language (NL) requirements for aerospace systems to formalism, the method comprising:

• receiving, as user input, an NL expression associated with a requirement for a landing gear system of an aircraft;
• using an LLM to:
• revise the NL expression to remove words that are determined by the LLM to be unnecessary to the NL expression;
• replace at least one word in the revised NL expression with a lifted atomic entity to generate an underlying sentence structure of the revised NL expression;
• translate the underlying sentence structure to a lifted logical formula;
• and
• ground the lifted logical formula by replacing each lifted atomic entity with a core concept such that multiple references to the same core concept are mapped to the same grounded atomic entity;
• parsing the grounded logical formula into an abstract syntax tree (AST) representation;
• using the LLM to convert the NL expression into a formalized requirement for the landing gear system of the aircraft based at least in part on the grounded logical formula;
• and
• controlling the landing gear system of the aircraft based on the formalized requirement.

A non-transitory computer-readable storage medium in which is stored instructions for causing a processor to execute a processor-implemented method for using a large language model (LLM) to convert natural language (NL) requirements for aerospace systems to formalism, the method comprising:

• receiving, as user input at the processor, an NL expression associated with a requirement for an aerospace system;
• using an LLM to:
• revise the NL expression to remove words that are determined by the LLM to be unnecessary to the NL expression;
• replace at least one word in the revised NL expression with a lifted atomic entity to generate an underlying sentence structure of the revised NL expression;
• translate the underlying sentence structure to a lifted logical formula;
• and
• ground the lifted logical formula by replacing each lifted atomic entity with a core concept such that multiple references to the same core concept are mapped to the same grounded atomic entity;
• parsing the grounded logical formula into an abstract syntax tree (AST) representation;
• using the LLM to reverse translate the AST representation into a NL representation of the grounded logical formula and to convert the NL expression into a formalized requirement for the aerospace system based at least in part on the grounded logical formula;
• and
• controlling the aerospace system based on the formalized requirement.