Improvement of the Log Pattern Extracting Algorithm Using Text Similarity
Yining Zhao, Xiaodong Wang, Haili Xiao, Xuebin Chi · 2018
Log processing and analysis are key jobs that can obtain information about system status and alerts, and log classification is the fundamental step. We have previously proposed the LARGE system and the algorithm of log pattern extraction to transfer system logs into a set of log pattern to help system maintainers find messages hidden in logs. But the proposed algorithm has the limitation that it cannot deal with log length differences. In this paper we will introduce the way of improving log pattern extracting algorithm by involving the idea of text similarity. Experiment results prove that the new algorithm shows obvious improvement than the original one. We also illustrate a modified log comparing model that fits real time log processing systems.