Automatic Program Code Generation from Documents Using Semantic Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automatic program code generation devices can only process conversational sentences and struggle to generate program code from thousands of sentences in documents like design sheets, manuals, and specifications, leading to increased workload and delays in operation automation.
Innovation Solution
An automatic program code generation system that includes text data extraction, semantic content searching, and code extraction using morphological analysis and neural networks to associate text data with relevant program code syntax, enabling the generation of program code from documents without manual intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If automatic program code generation is implemented using conventional intent-based methods, then conversational sentences can be processed, but documents with thousands of sentences cannot be automatically program-coded
Solution Approach 1:
The patent segments the document processing task into multiple stages: text extraction, morphological analysis to extract verbs/nouns/case components, semantic content matching against a database, and program code generation. This segmentation allows the system to handle complex documents by breaking them down into manageable processing steps, thereby enabling automatic coding of thousands of sentences while maintaining high adaptability across different document types.
2Reliability
If manual program code generation is performed for each new operation, then program accuracy can be maintained, but workload increases and operation delays occur
Solution Approach 1:
The patent implements preliminary action by pre-building a comprehensive database of semantic contents and their corresponding program code basic syntax before actual document processing. This pre-prepared knowledge base allows the system to quickly match and generate accurate program code during operation, eliminating the need for manual coding while maintaining reliability through the structured semantic matching process.
3Adaptability or versatility
If program code is rewritten manually for each operational change, then adaptability to new operations is achieved, but work burden increases enormously
Solution Approach 1:
The patent incorporates feedback mechanisms where the system continuously learns from processed documents and operational outcomes. The semantic content database is updated and refined based on accumulated processing experience, allowing the system to improve its program code generation accuracy over time. This feedback loop enables the system to adapt to new operations automatically without manual intervention, reducing both work burden and time loss while maintaining operational flexibility.
Data Source
AI summary
Text data is extracted from a document. Referring to a first trained model in which text data is associated with a semantic content with an association degree, a semantic content highly relevant to the text data is searched. Referring to a second trained model in which the semantic content is associated with program code basic syntax with an association degree, highly relevant program code basic syntax is extracted based on the semantic content.