Machine Learning Model for Code Logic Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Software enhancements often result in software defects due to human error during the manual development of target code for source code, particularly when upgrading from one version or technology to another.
Innovation Solution
A system utilizing machine learning techniques for continuous cognitive code logic detection and prediction, which receives source and target code scripts, generates a training dataset, trains a machine learning model, and deploys it to predict accurate target code scripts, thereby reducing human error and defects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual development of target code for source code is performed during software enhancement, then developers can create software enhancements, but software defects increase due to human error
Solution Approach 1:
The patent uses machine learning models to copy and generate target code scripts from source code scripts, replacing manual human coding. The system trains models on historical code transformations and uses them to automatically generate enhanced code, thereby maintaining productivity while eliminating human errors that cause defects.
Solution Approach 2:
The patent replaces the mechanical process of manual code writing with an automated machine learning-based system. The machine learning model processes source code inputs and generates target code outputs automatically, substituting human cognitive work with computational algorithms that can consistently produce defect-free code.
2Measurement precision
If machine learning model is trained on comprehensive code datasets, then prediction accuracy improves, but system complexity increases
Solution Approach 1:
The patent segments the code transformation task into multiple manageable components: data collection, data preprocessing, model training, and model deployment. This segmentation allows the system to handle comprehensive datasets systematically while maintaining manageable complexity through modular architecture and staged implementation.
Solution Approach 2:
The patent performs preliminary actions by pre-processing and organizing code datasets before training the machine learning models. Historical code transformations are collected, cleaned, and structured in advance, which simplifies the actual training process and improves prediction accuracy without increasing operational complexity during deployment.
Data Source
AI summary
Systems, computer program products, and methods are described herein for continuous cognitive code logic detection and prediction using machine learning techniques. The present invention is configured to receive, from a user input device, source code scripts and target code scripts for functional code logic components of a full stack, wherein the source code scripts and the target code scripts are associated with one or more tiers; generate a training dataset based on at least the source code scripts, the target code scripts, and the functional code logic components of the full stack; train, using a machine learning algorithm, a machine learning model using the training dataset; determine a prediction accuracy associated with the machine learning model; determine that the prediction accuracy is greater than a predetermined threshold; and deploy the machine learning model on unseen source code scripts.