System and Method for Iterative Skill-Gap Detection, Targeted Content Elicitation, and Automated Document Optimization Using Large Language Models

A system for generating an optimized document, the system comprising:

• at least one processor;
• at least one memory storing instructions that, when executed by the at least one processor, cause the system to:
• receive source document data comprising at least one content section and a target specification;
• analyze the target specification relative to the source document data using a large language model to identify at least one underrepresented requirement element having a coverage score below a configurable threshold;
• generate, via the large language model operating under a structured prompt template, at least one targeted elicitation query configured to extract from a user information regarding the underrepresented requirement element;
• receive from the user, via a user interface, a response to the elicitation query and a section identifier;
• transform the response into at least one structured content item via the large language model operating under a constrained generation prompt enforcing professional formatting constraints;
• compute a relevance score for the structured content item by comparison to the target specification;
• incorporate the structured content item into the content section identified by the section identifier to produce an optimized document;
• and
• responsive to the relevance score being below a threshold, generate at least one additional elicitation query for the same underrepresented requirement element and repeat the receive, transform, compute, and incorporate steps.

A non-transitory computer-readable medium storing instructions that, when executed by at least one processor, cause the at least one processor to perform operations comprising:

• receiving source document data and a target specification;
• identifying, using a large language model, at least one underrepresented requirement element in the source document data relative to the target specification, wherein the underrepresented requirement element has a coverage score below a configurable threshold;
• generating, using the large language model operating under a structured prompt template parameterized by the underrepresented requirement element, at least one targeted elicitation query;
• receiving a user response to the targeted elicitation query and a section identifier;
• transforming the user response into a structured content item using the large language model under a constrained generation prompt enforcing professional formatting constraints;
• computing a relevance score for the structured content item relative to the target specification;
• incorporating the structured content item into the source document data at the section identified by the section identifier;
• and
• iteratively generating additional elicitation queries and repeating the transforming and computing steps responsive to relevance scores below a threshold until a convergence criterion is satisfied.