Log representation learning for automated system maintenance

The method transforms system logs into structured templates, trains a language model, and detects anomalies to improve the efficiency of diagnosing system failures in cloud systems, addressing the challenge of vast unstructured data.

WO2025064274A1 Publication Date: 2025-03-27NEC LABORATORIES AMERICA INC
5 Cites 0 Cited by

Patent Information

Application Number
PCT/US2024/046125
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2024-09-10
Filing Date
2024-09-11
Publication Date
2025-03-27

AI Technical Summary

Technical Problem

Current cloud systems face challenges in efficiently diagnosing system failures due to the vast amount of unstructured data generated, which makes it difficult to identify vulnerabilities and requires significant time and resources.

Method used

A computer-implemented method and system for log representation learning that transforms system logs into log templates using an optimized parser, tokenizes these templates into log template tokens, trains a language model with deep learning, detects anomalies, and performs corrective actions based on the detected anomalies.

Benefits of technology

This approach enables the extraction of high-quality representations from system logs, facilitating the diagnosis of failures in cloud and microservice systems, thereby improving the efficiency and accuracy of automated system maintenance.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

Systems and methods for log representation learning for automated system maintenance. An optimized parser can transform (110) collected system logs into log templates. A tokenizer can tokenize (120) the log templates partitioned into time windows to obtain log template tokens. The log template tokens can train (130) a language model (LM) with deep learning to obtain a trained LM. The trained LM can detect (140) anomalies from system logs to obtain detected anomalies. A corrective action can be performed (150) on a monitored entity based on the detected anomalies.
Need to check novelty before this filing date? Find Prior Art