Work Item Impact Prediction for Code Repository Files
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The process of updating software and navigating codebases for new developers is inefficient due to the time-consuming task of mapping natural-language terms to code constructs, leading to significant inefficiencies in software development and onboarding new hires.
Innovation Solution
Implementing a computer-implemented method using a trained machine learning model to determine correlation strengths between work items and files, enabling automatic prediction of which files will be modified, and applying machine learning models to generate representations of file modifications based on work items.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If natural-language terms are manually mapped to code constructs, then accuracy of file identification is improved, but time consumption increases significantly
Solution Approach 1:
The system performs preliminary actions by pre-processing work item descriptions to extract topics and pre-calculating correlation strengths between topics and files before the actual update process. This allows the machine learning model to quickly predict affected files without manual mapping during execution, thereby reducing time consumption while maintaining accuracy.
Solution Approach 2:
The patent replaces the manual mechanical process of mapping natural-language terms to code constructs with an automated machine learning system. The ML model automatically analyzes work item descriptions, extracts relevant topics, and predicts affected files based on learned correlations from training data, eliminating the need for manual term-to-code mapping while maintaining or improving accuracy.
2Reliability
If codebase traversal is performed to identify affected files, then completeness of impact analysis is improved, but processing speed deteriorates
Solution Approach 1:
The system performs preliminary analysis by extracting topics from work item descriptions and pre-calculating correlation strengths between these topics and files in the codebase during training. This preliminary preparation enables the ML model to quickly predict affected files without performing complete codebase traversal during actual updates, thus maintaining completeness while improving processing speed.
Solution Approach 2:
The patent creates a conceptual copy or representation of the codebase structure and file relationships that can be efficiently queried by the machine learning model. Instead of traversing the actual codebase during updates, the system uses this pre-built representation to quickly identify affected files, maintaining completeness of analysis while dramatically improving processing speed.
3Measurement precision
If machine learning models are trained on work item data, then prediction accuracy of file modifications is improved, but initial setup time and computational resources increase
Solution Approach 1:
The system performs the computationally intensive training of machine learning models as a preliminary action during system initialization or periodic updates. Once trained, the models can quickly predict affected files with high accuracy during actual work item processing. This separates the heavy initial setup cost from the operational phase, making the trade-off acceptable for ongoing productivity gains.
Data Source
AI summary
A computer-implemented method, according to one approach, includes: receiving a new work item, and extracting topics from the new work item. A trained machine learning model is used to determine a first set of values representing correlation strengths between the new work item and the topics. Moreover, the first set of values are compared to a second set of values, where the second set of values represents correlation strengths between the topics and multiple files. A third set of values representing correlation strengths between the new work item and the multiple files is also generated, and output.