Streaming online log parsing method based on heuristic rule

To address the issues of inaccurate parsing and unstable performance in existing log parsing methods, a streaming online log parsing method based on heuristic rules, known as heuristic regex tree (HRTree), was proposed. Based on the drain method of parsing the structure tree, heuristic rules were introduced to split and construct the log, and some construction methods of the parse structure tree were optimized, so as to solve the problems of over fitting of log parameters and unstable parsing results of different system logs. Not only the classification of parsing results was accurate, but also the parameter content recognition was accurate. A large number of experimental results demonstrate that the proposed HRTree parsing method shows more than 90% parsing accuracy on different system logs.