Network Status Classification from Mixed Text and Numeric Logs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for analyzing log data in communication networks fail to efficiently extract useful information due to the variability in formatting and content across different manufacturers, leading to difficulties in identifying system failures and optimizing network performance.
Innovation Solution
A method that transforms both textual and numerical elements of log data into respective vector spaces and clusters them to train a classifier model, allowing for a more comprehensive analysis of network status by embedding and clustering textual and numerical elements into a single embedded vector representing a wireless device session.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual analysis of log events is used, then extraction of necessary information is possible, but the process is time consuming and does not scale efficiently
Solution Approach 1:
The patent replaces manual mechanical analysis with automated machine learning-based text mining and pattern recognition systems. The system automatically extracts information from log events using trained models, eliminating the need for manual analysis while maintaining high accuracy and enabling scalable processing of large volumes of log data.
2Ease of operation
If keyword searching based on experience and domain-specific knowledge is used, then evaluation of log data is possible, but the searching is focused on a limited subset time range and can omit critical information
Solution Approach 1:
The patent transitions from traditional one-dimensional keyword searching to multi-dimensional analysis by incorporating temporal, contextual, and relational dimensions. The system analyzes log events across multiple time ranges simultaneously, considers contextual relationships between events, and uses vector space models to capture semantic meanings beyond simple keyword matching, thereby retrieving comprehensive information without omitting critical details.
3Adaptability or versatility
If known keywords repository is continuously updated to match system hardware and software upgrades, then relevance of log searching is maintained, but the complexity of maintenance increases
Solution Approach 1:
The patent implements self-updating mechanisms where the system automatically learns and adapts to new hardware and software configurations through continuous training on incoming log data. The machine learning models automatically adjust to new keywords, patterns, and event types without requiring manual intervention, making the system self-maintaining and adaptable to system upgrades without increasing operational complexity.
4Extent of automation
If existing automated log data processing techniques focusing on textual content are used, then text analysis is possible, but important information from measurement variables is omitted
Solution Approach 1:
The patent merges previously separate processing streams for textual log events and numerical measurement variables into a unified analysis framework. The system simultaneously processes both text and numeric data, integrating them through feature fusion techniques that combine semantic understanding with quantitative analysis, thereby preserving all important information from both sources rather than losing measurement variable data.
Data Source
AI summary
A method includes extracting, from first log data relating to operations of wireless devices and/or nodes in the communication network, textual elements and numerical elements; transforming the textual elements to a first vector space to determine respective textual element vectors; transforming the numerical elements to a second vector space to determine respective numerical element vectors; embedding and clustering the textual element vectors and the numerical element vectors to determine clusters of embedded vectors, wherein the embedding includes, for a plurality of wireless device sessions, embedding at least one textual element vector and at least one numerical element vector into a single embedded vector representing the particular wireless device session; and training a classifier model to determine a network status from second log data, wherein the classifier model is trained using the plurality of clusters of embedded vectors.