Software Issue Triage With Impact Classification and AI Descriptions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional software application issue detection systems fail to efficiently triage and provide understandable descriptions of issues, overwhelming users with large numbers of detected issues and requiring manual analysis to understand their impact and effect on user experience.
Innovation Solution
Techniques for automatically triaging issues by determining impact levels using machine learning and generating natural language descriptions, focusing on high-impact issues and reducing computational resources by limiting language model processing to high-impact issues.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If all detected issues are analyzed manually, then complete issue understanding is achieved, but time consumption and resource requirements increase significantly
Solution Approach 1:
The system segments issues by determining impact levels (e.g., high, medium, low) and separates them into different categories. This allows manual analysis to focus only on high-impact issues while automated processing handles lower-impact ones, reducing overall analysis time while maintaining understanding accuracy for critical issues.
Solution Approach 2:
The system generates natural language descriptions of issues automatically using language models, allowing the system to self-describe issues without requiring manual analysis. This self-service capability provides sufficient understanding for many issues, reserving manual analysis only for cases where deeper insight is needed.
2Loss of information
If language model processing is applied to all issues, then comprehensive descriptions are generated, but computational resources are wasted on low-impact issues
Solution Approach 1:
The system applies different processing quality to different issues based on their impact level. High-impact issues receive comprehensive natural language descriptions through language model processing, while low-impact issues receive simpler or automated descriptions. This local quality differentiation maintains description quality where needed while reducing computational resource consumption overall.
3Reliability
If manual analysis is performed on all detected issues, then complete issue triage is achieved, but productivity decreases due to the large volume of issues
Solution Approach 1:
The system segments the large volume of detected issues into different impact level categories (high, medium, low). This segmentation enables automated high-throughput processing of low-impact issues while directing manual analysis resources to high-impact issues, thereby maintaining triage accuracy for critical issues while significantly improving overall processing throughput and productivity.
Data Source
AI summary
Described herein are techniques that use machine learning to triage issues by classifying the issues into impact levels. Described herein are also techniques for generating a natural language description of issues that occur in sessions of a software application. The techniques collect data during sessions in which a user is interacting with the software application. The techniques process the data collected during the sessions using a language model to obtain natural language descriptions of issues that occur in the sessions.