Log Sentiment Analysis Using Domain-Specific Lexicon
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing log analysis methods struggle to accurately analyze sentiment in log messages due to the use of generic vocabulary, leading to inaccurate sentiment values and a lack of visibility into subtle abnormalities in system operations, which hinders timely remedial actions and root cause diagnosis.
Innovation Solution
A cognitive alert system that creates a log data-specific lexicon with weighted sentiment scores, processes log messages to extract relevant data entities, and uses machine learning to classify messages based on sentiment and priority, generating alerts with predefined key performance indicators to inform users of critical issues in real-time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If generic vocabulary is used for log analysis, then the analysis process is simple, but the sentiment analysis accuracy is poor
Solution Approach 1:
The patent applies local quality by creating a specialized lexicon tailored specifically for log data analysis rather than using generic vocabulary. The lexicon is customized to capture domain-specific sentiment patterns in log messages, improving measurement precision for sentiment analysis while maintaining manageable complexity through targeted word selection and weighting mechanisms.
Solution Approach 2:
The patent implements parameter changes by introducing weighted sentiment scores for different words in the log-specific lexicon. Instead of treating all words equally, the system assigns varying weights based on their importance and sentiment strength in log contexts, thereby enhancing analysis accuracy without proportionally increasing system complexity.
2Reliability
If generic log analysis is performed, then the system operation is easy to implement, but subtle abnormalities are not detected
Solution Approach 1:
The patent applies preliminary action by pre-processing log data to extract relevant features and pre-configuring the sentiment lexicon with weighted scores before actual analysis. This preparation work is done in advance, allowing the system to detect subtle abnormalities more reliably during operation without requiring complex real-time processing, thus maintaining ease of operation while improving reliability.
Solution Approach 2:
The patent replaces simple mechanical log matching with a more sophisticated sentiment analysis mechanism that uses weighted scoring and classification algorithms. This substitution enables the system to detect subtle abnormalities by analyzing sentiment patterns rather than relying on exact keyword matches, improving reliability while the automated nature of the system keeps operational complexity manageable.
3Loss of information
If detailed sentiment analysis is performed on all log messages, then the visibility into system operations is improved, but the processing time increases
Solution Approach 1:
The patent applies the extraction principle by selectively extracting only the most relevant features and words from log messages for sentiment analysis. Instead of analyzing every aspect of each log message in detail, the system identifies and focuses on key sentiment-bearing elements, thereby maintaining high information visibility while reducing overall processing time by ignoring less relevant data.
Solution Approach 2:
The patent implements partial action by performing detailed sentiment analysis only on log messages that meet certain criteria or contain specific keywords indicating potential issues. Not all log messages undergo full detailed analysis - the system applies varying levels of analysis depth based on message characteristics, thus improving information visibility for critical messages while avoiding unnecessary processing time expenditure on routine logs.
Data Source
AI summary
Methods, computer program products, and systems are presented. The methods include, for instance: creating a log data specific lexicon based on log data samples, each word in the log data specific lexicon corresponding to a weighted sentiment score with a binary polarity. A log message is obtained, and a sentiment value of the log message is assigned based on respective weighted sentiment scores of words appearing in the log message. The log message is classified for a class indicating an issue the log message addresses. An alert type for the log message is determined based on the sentiment value, the class, and a priority of the log message, and a system alert including a set of key performance indicators according to the alert type is produced to a user.