Automated Bug Classification Model Using Source Code Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing bug tracking tools are inadequate for real-time detection of defective code, often miss bugs due to hardware limitations and are not dynamically integrated into the software development lifecycle, leading to increased bug counts over time.
Innovation Solution
An automated classification system that receives bug reports from tools like BUGZILLA, clusters source code, and trains a classification model to detect bug presence using supervised and unsupervised learning, allowing continuous integration and real-time detection of bugs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing bug tracking tools are used, then bug tracking is possible, but real-time detection of defective code is inadequate and bugs are missed due to hardware limitations
Solution Approach 1:
The patent replaces traditional mechanical bug tracking tools with an automated machine learning classification system that processes source code directly. The system uses supervised learning models to automatically detect and classify bugs, eliminating the need for manual tracking and overcoming hardware limitations of traditional tools.
Solution Approach 2:
The system enables source code to be automatically analyzed and classified for bugs without requiring external intervention. The machine learning model processes code independently, automatically identifying defective portions and generating bug reports, making the bug detection process self-sufficient.
2Reliability
If bug tracking tools are not dynamically integrated into the software development lifecycle, then integration is simpler, but bug counts increase over time
Solution Approach 1:
The bug detection system is designed to be universally applicable across different stages of the software development lifecycle. It can integrate with various development workflows and continuously process code changes, making it adaptable to different integration scenarios while maintaining consistent bug detection capability.
Solution Approach 2:
The system dynamically adapts to changes in the software development lifecycle by continuously processing new code and updating bug classifications. The machine learning model can be retrained with new data, allowing the system to evolve and improve its detection capability as the project progresses.
3Measurement precision
If manual bug tracking is used, then implementation is simpler, but detection precision and reliability decrease
Solution Approach 1:
The patent replaces manual bug tracking with automated machine learning classification. The system uses supervised learning models to automatically analyze source code, identify bugs, and classify them by type, achieving high detection precision without manual intervention.
Solution Approach 2:
The system transforms the bug detection process by changing from manual parameter assessment to automated feature extraction and classification. The machine learning model analyzes multiple code parameters simultaneously, identifying patterns that indicate bugs with high precision.
Data Source
AI summary
Systems and methods are described for automated classification of defective code from bug tracking tool data. An example method includes receiving a plurality of datasets representing a plurality of bug reports from a bug tracking application. Each dataset may be generated by vectorizing and clustering a source code associated with a respective bug report represented by the dataset. Each dataset may comprise a plurality of classes. At least one class of each dataset may indicate at least one known bug. For each dataset of the plurality of datasets, a respective supervised feature vector may be generated. Each supervised feature vector may be associated with an index of the at least one class with the at least one known bug. Using the supervised feature vectors, a classification model is trained to detect a new bug presence in a new source code.