MACLog: Multi-Algorithm Collaborative Log Parsing Approach

Log parsing approach is capable of converting the raw log into the structured data, which is an important foundation for analyzing logs. Due to the diversity of log types, there is a diversity of log parsing approachs. Furthermore, there is no an approach that can handle all logs well. Therefore, this paper proposes a multi algorithm collaborative log parsing approach, which we call MACLog. MACLog utilizes log recognition algorithms to identify the types of logs to be processed, and generates collaborative strategies based on a knowledge graph for log anomaly detection, thereby implementing multi algorithm collaboration. We use the logs provided by LogHub as the experimental dataset. The experimental results indicate that MACLog can effectively improve the accuracy of log parsing.