Chat Prompt Control for Complete Required Information Gathering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Large language models (LLM) sometimes provide inappropriate answers, such as vague or irrelevant responses, when necessary information is not input for generating an appropriate answer to a prompt.
Innovation Solution
A chat control system that includes determination and generation processing to identify and generate required information, output prompts to the LLM, and receive answers to ensure all necessary information is gathered before generating a chat message, thereby preventing inappropriate responses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the LLM is provided with minimal input information to maintain interactive efficiency, then the response time is reduced, but the answer quality deteriorates (vague or irrelevant answers)
Solution Approach 1:
The system performs preliminary determination of required information and generates missing information before constructing the final prompt. This advance preparation ensures that when the LLM receives the prompt, all necessary information is already available, preventing vague or irrelevant answers while maintaining efficient interaction.
Solution Approach 2:
The system introduces an intermediate processing layer between user input and LLM processing. This intermediary component determines required information, generates missing data through information generation processing, and assembles complete prompts, thereby ensuring answer quality without sacrificing interactive efficiency.
2Reliability
If the system waits for all required information to be generated before outputting the prompt, then answer quality improves, but the response time increases
Solution Approach 1:
The system performs information determination and generation as preliminary actions before prompt construction. By completing these tasks in advance and only then outputting the prompt when all required information is available, the system ensures appropriate answers without excessive delays during the actual interaction.
Solution Approach 2:
The system implements a feedback mechanism that monitors whether all required information has been generated. The prompt is output only when the determination confirms complete information availability, creating a controlled waiting process that ensures answer quality while minimizing unnecessary delays.
3Loss of information
If the system generates multiple pieces of required information through iterative chat messages, then information completeness improves, but the number of interaction turns increases
Solution Approach 1:
The system determines all required information in advance and performs generation processing for missing information before the final prompt is constructed. This preliminary completion of information gathering reduces the need for multiple iterative chat turns, achieving information completeness more efficiently.
Solution Approach 2:
The system introduces an intermediary information generation component that automatically produces missing required information based on the chat message and determination results. This intermediary processing reduces the need for additional user interactions, completing information gathering in fewer turns while maintaining completeness.
Data Source
AI summary
A determination and generation processing execution module executes determination and generation processing for each piece of non-generated required information. A determination and generation control module causes, when at least one of the pieces of non-generated required information is not generated based on a chat message, in response to reception of a next chat message from a user terminal, the determination and generation processing execution module to execute the determination and generation processing for the next chat message. A prompt generation module generates, when all of the pieces of non-generated required information are generated based on the chat message, a prompt based on the plurality of pieces of required information.