Framework for structured Prompt Building for a generative language model
Topics: Chunk Relevance, LLM Readability, LLMO / GEO, Microsoft, Probably in use, Prompt Engineering, Query Fan Out, Search Query Processing
This Microsoft patent describes a system and method for creating structured prompts to enhance user interaction with Large Language Models (LLMs). The system uses a computer-implemented method to dynamically refine an initial user query into a detailed prompt by presenting selectable intent indicators and offering in-line word and phrase recommendations. By guiding the user through the refinement process, this framework simplifies engagement with advanced LLM capabilities, ultimately improving the accuracy and relevance of the LLM’s responses and enhancing the overall search experience.
