Cloud Template Graph Matching for Accurate User-Guided Creation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing large language models (LLM) struggle to generate accurate templates for cloud development due to incomplete user input, leading to parameter inconsistencies and resource definition errors, which require significant manual correction.
Innovation Solution
A template creation support system that converts user inputs and existing templates into graphs representing service dependencies, allowing for the search and presentation of similar templates with higher accuracy by considering service relationships, thereby guiding users in completing necessary input.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If LLM generates template based on user input, then template creation speed is improved, but template accuracy deteriorates due to incomplete user input
Solution Approach 1:
The system performs preliminary actions by converting existing templates into graph representations in advance and storing them. When a user requests template generation, the system compares the user's input graph against these pre-processed templates to identify setting items that need to be added or corrected, thereby improving template accuracy without sacrificing generation speed
Solution Approach 2:
The system implements feedback by analyzing the user's input graph, comparing it with existing templates, and identifying specific setting items that are missing or incorrect. This feedback mechanism guides users to supplement incomplete information, ensuring that the final generated template achieves high accuracy while maintaining fast generation throughput
2Manufacturing precision
If user inputs all necessary setting items, then template accuracy is improved, but ease of operation deteriorates due to complex input requirements
Solution Approach 1:
The system applies self-service by automatically analyzing the user's simple input graph and autonomously identifying which setting items are missing or incorrect by comparing against existing templates. The system then presents this analysis to the user, allowing them to easily supplement only the necessary information without needing to manually input all setting items from scratch
Solution Approach 2:
The system uses the graph representation as an intermediary between the user's simple natural language input and the complex template structure. By converting both user input and existing templates into graphs, the system can automatically compare and identify differences, guiding users to provide only the minimal necessary information to achieve accurate template generation
3Manufacturing precision
If manual correction is performed on LLM-generated templates, then template accuracy is improved, but loss of time increases due to extensive corrections needed
Solution Approach 1:
The system performs preliminary actions by pre-converting existing templates into graph representations and storing them with their associated setting items. When generating a new template, the system immediately compares the user's input graph against these pre-processed templates to identify necessary corrections, dramatically reducing the time required for manual verification and adjustment while ensuring high template accuracy
Data Source
AI summary
Provided is a template creation support system capable of supporting creation of a template of a flow close to a system configuration desired to be realized by a user. There are provided a first conversion unit that converts each template of a template group into a graph including a node indicating a service and an edge indicating a relationship between services based on a dependency relationship between set resources, a second conversion unit that converts an input including information indicating a system configuration desired by a user and an operation of each component into a graph including a node indicating a service and an edge indicating a relationship between services, a search unit that searches a template of a graph close to the graph converted by the second conversion unit from among the graphs converted by the first conversion unit from the template group, and an output unit that outputs a setting item set to the template searched by the search unit.