EvoParser: An Evolutionary Approach to Log Parsing
Jiří Setinský, Martin Žádník · 2025
In the domain of network operations and management, efficient log parsing is crucial for processing the high volumes of log data necessary to monitor the reliability and performance of the operated infrastructures and services. This paper introduces a novel method for creating highly efficient log parsers. Our method employs a genetic algorithm to evolve optimized log parsing graphs that deliver high-quality parsing results. Experimental evaluations demonstrate that our approach achieves results comparable to or better than state-of-the-art techniques while maintaining low complexity during parsing. These characteristics make it particularly suitable for real-time applications in high-volume data scenarios.