Conversion Program Search System for Schema Integration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems face challenges in schema integration due to differences in attribute value description formats, requiring manual conversion rules and lengthy processing times to find suitable conversion modules, which limits the reusability of conversion modules and efficiency in data connection problems.
Innovation Solution
A conversion program search system that generates signatures to identify suitable conversion paths by comparing input and output data formats, selecting candidate conversion programs based on similarity, and calculating evaluation values to efficiently match conversion-source and post-conversion data formats, thereby reducing processing time and increasing reusability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual conversion rules are created to handle schema integration, then conversion accuracy is improved, but processing time and system complexity increase
Solution Approach 1:
The system performs preliminary actions by automatically generating conversion rules through machine learning models before actual data conversion is needed. The model is trained on historical conversion data and conversion examples to pre-establish conversion patterns, so when new schema integration is needed, the system can quickly apply pre-learned rules rather than manually creating them each time.
Solution Approach 2:
The system enables self-service by allowing the machine learning model to automatically generate and refine conversion rules without human intervention. The model learns from conversion examples and continuously improves its rule generation capability, making the system self-improving and reducing dependency on manual rule creation while maintaining high accuracy.
2Adaptability or versatility
If comprehensive conversion rules are manually created, then conversion coverage is improved, but ease of operation deteriorates
Solution Approach 1:
The system enables self-service by allowing the machine learning model to automatically generate and refine conversion rules without human intervention. The model learns from conversion examples and continuously improves its rule generation capability, making the system self-improving and reducing dependency on manual rule creation while maintaining high accuracy.
Solution Approach 2:
The system changes parameters by using probabilistic models and confidence scores to dynamically adjust conversion rule application. Instead of rigid manual rules, the system evaluates multiple potential conversions and selects the most appropriate one based on learned patterns and confidence levels, improving adaptability while maintaining ease of operation through automated decision-making.
3Adaptability or versatility
If multiple conversion programs are searched manually, then reusability of conversion modules is improved, but productivity decreases
Solution Approach 1:
The system performs preliminary actions by pre-processing conversion examples and training data to extract reusable conversion patterns. The machine learning model is trained in advance on historical conversion data, so when new conversion tasks arise, the system can quickly query and apply relevant pre-learned patterns rather than searching through multiple conversion programs manually each time.
Solution Approach 2:
The system replaces the mechanical process of manually searching and selecting conversion programs with an automated machine learning-based system. The model automatically identifies relevant conversion patterns from training data and applies them to new tasks, substituting manual mechanical search operations with intelligent automated pattern recognition and application.
Data Source
AI summary
To provide a conversion program search system that can increase the reusability of conversion programs as well as can find a solution to a data connection problem in a short time. A first-signature generation means 71 generates a first signature indicating, in a binary manner, whether one piece of data meets each of predetermined multiple description formats. A second-signature generation means 72 generates a second signature of a conversion program, based on the difference between the first signatures of conversion-source data and post-conversion data. When a conversion path candidate generation means 74 generates a candidate for a solution to a data connection problem, an evaluation value calculation means 75 calculates the evaluation value of the candidate using the second signature. A conversion path identification means 76 tries conversion paths to serve as a solution in the descending order of the evaluation values.