Natural Language App Generation With Structured Code Configuration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing low-code development platforms using large language models (LLM) for natural language processing lack clear disclosure of specific configurations for application development, limiting their effectiveness in generating application components and documentation based on user inputs.
Innovation Solution
A method and program that utilize a supervised learning approach to fine-tune a language model (LM) for generating application code and documentation from natural language commands, incorporating model objects, view template objects, and action objects to create user input and output components, and filter data based on user-defined conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a large language model (LLM) is used for natural language processing in low-code development, then the development process becomes simpler and more efficient, but the specific configuration for generating application components and documentation is not clearly disclosed
Solution Approach 1:
The patent segments the LLM's natural language processing capability into distinct functional modules: user input component generator, data management component generator, data visualization component generator, and documentation generator. Each module receives specific configuration parameters and generates corresponding application components, making the previously black-box process transparent and controllable.
Solution Approach 2:
The patent performs preliminary configuration by defining specific input parameters (user input type, data source, visualization type, etc.) before the LLM generates application components. This preliminary structuring of configuration information ensures that the generated components have the necessary details for immediate use, eliminating the need for post-generation configuration adjustments.
2Ease of operation
If natural language commands are used to generate application code, then ease of operation improves, but manufacturing precision of the generated code may deteriorate
Solution Approach 1:
The patent implements a feedback mechanism where the LLM's generated code is validated against the original natural language command and predefined configuration parameters. If the generated code does not accurately reflect the user's intent or contains errors, the system provides feedback to the LLM for refinement, ensuring high precision in code generation while maintaining natural language input simplicity.
Solution Approach 2:
The patent transforms natural language commands into structured configuration parameters with specific data types and validation rules. By converting unstructured natural language into parameterized form, the system maintains the ease of natural language input while ensuring the precision required for accurate code generation through controlled parameter specifications.
Data Source
AI summary
A method for generating an application program according to an embodiment disclosed herein includes inputting a natural language command through a developer input interface from a developer who develops the application program, inputting the input natural language command into a predetermined language model, acquiring predetermined output information generated by the language model in response to the input of the natural language command, and generating a source code or an execution code of the application program using the acquired output information.