Error Log Standardization Using Neural Labeling and Natural Language
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Error logs from different software applications, operating systems, and servers are formatted differently, making them difficult to interpret and analyze, especially for troubleshooting, due to the presence of machine-generated codes with occasional natural language components that are not sentences.
Innovation Solution
A predictive neural network is trained to convert unlabeled error logs into a standardized and labeled format by predicting appropriate labels for keywords and descriptive natural language text, using a training data set and keyword extraction algorithms to generate metadata that summarizes the underlying performance issues.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If error logs are captured in different formats for different applications and operating systems, then the error logs can capture application-specific information, but the error logs become difficult to interpret and analyze
Solution Approach 1:
The patent introduces a natural language generation component as an intermediary that translates machine-generated error codes into human-readable natural language sentences. This mediator layer converts the disparate format types into a unified, interpretable format without losing application-specific information, thereby resolving the contradiction between format adaptability and interpretability
Solution Approach 2:
The system changes the parameter of error log representation from machine-coded format to natural language format. By transforming the output parameter (error message format) while maintaining the input parameter (error code information), the system achieves both format adaptability and improved interpretability
2Extent of automation
If machine-generated codes are used in error logs, then the error logs can be generated automatically by systems, but the error logs contain components that are not sentences and are difficult to understand
Solution Approach 1:
The patent replaces the mechanical system of machine-generated codes with a linguistic system that produces natural language sentences. The neural network model substitutes traditional error message formatting with automated natural language generation, maintaining automation while improving readability through sentence-based output
3Loss of information
If error logs capture all available information, then the error logs contain comprehensive troubleshooting data, but the error logs require special administrative rights and are complex to access
Solution Approach 1:
The patent segments the error log information into structured natural language sentences with identified keywords and entities. This segmentation organizes comprehensive error information into accessible, labeled components that can be easily queried and analyzed without requiring complex administrative access procedures
Data Source
AI summary
Embodiments of the invention are directed to computer-implemented methods of labeling unlabeled electronic information. In a non-limiting embodiment of the invention, the computer-implemented method includes receiving, using a processor system, an unlabeled error log (EL) having an EL format. A set of unlabeled EL keywords are extracted from the unlabeled EL. For each unlabeled EL keyword, the processor system uses the unlabeled EL keyword to extract an electronic document (ED) sentence from an ED based on a determination that the ED sentence is relevant to the unlabeled EL keyword. One or more ED keywords are extracted from the ED sentence. A deep neural network of the processor system is used to predict an ED sentence label for the ED sentence, an ED keyword label for the ED keyword, and an EL keyword label for the unlabeled EL keyword.