ML-Generated Log Search Programs for Accurate Pattern Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The manual extraction of useful parts from large-sized log data in information processing systems is burdensome and inefficient, often resulting in low accuracy due to varying formats and the need for extensive programming and string matching, which is not effective for log data with many situation-dependent character strings.
Innovation Solution
A computer-based system using a trained machine learning model generates a search program to identify partial log data patterns, reducing the burden of manual programming and improving extraction efficiency by creating a search program capable of finding similar patterns in log data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual extraction methods are used to identify useful parts from log data, then workers can extract failure detection or recovery information, but the process becomes burdensome and inefficient with low accuracy due to varying formats and extensive programming requirements
Solution Approach 1:
The patent replaces manual mechanical extraction processes with an automated machine learning-based system. The machine learning model automatically identifies and extracts useful log data patterns without requiring manual programming or string matching, thereby improving both accuracy and efficiency simultaneously.
Solution Approach 2:
The machine learning model enables the log data extraction process to be self-service oriented. The system automatically learns from log data patterns and performs extraction without continuous human intervention, allowing workers to obtain accurate results without bearing the burden of manual programming and format analysis.
2Adaptability or versatility
If string matching and programming are used to extract log data, then specific patterns can be identified, but the approach is not effective for log data with many situation-dependent character strings and requires extensive programming
Solution Approach 1:
The patent changes the fundamental parameter of pattern matching from fixed string matching to dynamic pattern recognition by the machine learning model. This allows the system to adapt to various log data formats and situation-dependent character strings without requiring complex programming for each format variation.
Solution Approach 2:
The machine learning model provides a universal solution that handles multiple log data formats and patterns through a single system. Instead of requiring separate programming for different log formats, the model generalizes pattern recognition across various formats, reducing programming complexity while maintaining versatility.
3Quantity of substance
If conventional extraction methods are used, then log data can be processed, but the workload is high and execution time is long
Solution Approach 1:
The machine learning model performs preliminary learning from log data patterns during training, so that during actual extraction, it can quickly identify and extract useful information without time-consuming manual analysis or complex string matching operations, thereby reducing execution time while processing large volumes of log data.
Data Source
AI summary
A computer acquires first partial log data extracted from first log data output by an information processing system. The computer generates, by entering the first partial log data to a trained machine learning model, a search program for searching second log data for second partial log data having a common pattern with the first partial log data by using the machine learning model.